Hi friends 👋,

Happy Friday and welcome to our 191st Weekly Dose of Optimism!

Packed Dose today - we had to double-up a few related stories to fit it all. We have everything from cancer-sniffing dogs to space-based energy lasers. Plus, the Sixers crushed the Celtics last night to force a Game 7 and it’s a perfect 50 and sunny heading to 62 here in New York City heading into the first May weekend of 2026… my god, life is good.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Create

Every morning, the kiddos and I mix two packets of Creatine + Electrolytes into water and drink it. This morning, we did Passionfruit. Delicious.

If you’ve been wondering why I look so strong, well-recovered despite Hyrox training, sharp, and hydrated, it’s gotta be the C+E. It’s not just me though. In just a month since launch, the product is already run-rating $20 million.

Each packet contains 5g creatine monohydrate, 800mg electrolytes, and 1000mg taurine. I These little packets, mixed into a glass of water, support hydration, improve recovery, increase energy, and boost cognition.

So if you want a great-tasting extra kick for your muscles and your brain, and you want to support long-time Dose writer and my brother Dan, get 30% off C+E here:

Get 30% Off Create C+E

(1) Dogs That Can Sniff Out Cancer

Tigerfeathers on Dognosis

Last year, waiting in the taco line at the Progress Conference, I started talking to the guy in front of me, Akash Kulgod, who told me that he was working on using dogs to detect cancer. His company is called Dognosis.

The idea is that dogs have an exquisite sense of smell but don’t speak human, so they train teams of beagles, labradors, and shelter dogs to sniff disease signatures in human breath samples (collected by having the patient breathe normally into a face mask for 10 minutes) while wearing EEG helmets, sensor suits, and being recorded on video so AI can read the dog's perceptual judgment as a digital signal.

I love dogs and I hate cancer, so I thought it was an awesome idea, but woof did it seem like a longshot.

Well, last week, the Journal of Clinical Oncology published the results of their Phase 2 study, covering 3,275 participants across six hospitals in Karnataka, the largest breath-based multi-cancer detection study ever run. Each sample was assessed independently by at least three dogs, and a Bayesian fusion model combined their judgments. And the results are…

Hold up. For context here, the leading blood-based Multi-Cancer Early Detection (MCED) test on the market, GRAIL's Galleri, runs ~$500–$1,000 per test and posts overall sensitivity of ~51.5% with stage-dependent performance ranging from 16.3% at Stage I to 90.1% at Stage IV. Meaning, today's gold standard MCED catches about 1 in 6 Stage I cancers. Early stage cancer detection is really hard, but it’s also where you have the best chance of just knocking it out.

Dognosis scored a 90.8% on sensitivity and 91.3% on specificity across seven cancer types (oral, breast, esophageal, cervical, lung, colorectal, prostate), which is excellent, but the insanely great part is that the sensitivity for early-stage (Stage I-II) disease held at 90.6%, showing basically no degradation versus late-stage detection. That is huge. Catching cancer early is the best cure we have.

In other words, today's leading blood test catches about 1 in 6 Stage I cancers; Dognosis catches 9 in 10. Galleri is a little more specific (fewer false positives) but Dognosis is meant to be radically cheaper, which is important in places like India where 80% of cancers are caught at Stage III or IV and roughly 1% of the population has ever been screened, and which is important to get more early screening generally. After the cheap test, we can figure out how to lower false positives at subsequent, more expensive stages.

The next step is real-world programs across multiple Indian states, with a U.S. study to follow.

This is an incredibly cool company showing very promising early results, made in India, meant for the world. For much, much more, our friend Rahul Sanghi at Tigerfeathers did a super in-depth interview with Akash that you should read.

Woof woof, woof. (Get fucked, cancer.)

(1b) Mayo Clinic AI helps detect pancreatic cancer up to 3 years before diagnosis

Man all the homies hate cancer and are figuring out how to detect it early.

Last week, we led with the six-year follow-up on BioNTech and Genentech's personalized mRNA pancreatic cancer vaccine, writing, “it’s sometimes called the ‘silent killer’ because symptoms don’t show up until very late, and by the time they do, only about 1 in 10 patients has a tumor that’s still operable.”

Early detection would help enormously, and it turns out, AI might be able to help. In a paper published in Gut, Mayo researchers showed that an AI model called REDMOD, run on routine abdominal CT scans that had already been read as normal, can detect pancreatic cancer up to three years before clinical diagnosis.

REDMOD showed 73% sensitivity vs. 39% for specialist radiologists looking at the exact same scans, at 88% specificity, with a median lead time of about 16 months. For cancers that eventually showed up more than two years later, REDMOD flagged 68% of them while the radiologists got 23%. The model is now being moved into a prospective trial called AI-PACED to test it in real care pathways for high-risk patients (e.g. people with new-onset diabetes, which is sometimes an early sign of pancreatic cancer).

Woof woof, woof.

(2) Zuck Commits $500M to Bio & Space Energy Lasers

Yesterday, in Scarce Assets, I wrote about Zuck’s purchase of the most expensive home in Miami history at $170 million, beating the previous, pathetically low record by $50 million. Later that afternoon, rumors emerged that Zuck was one of the people interested in buying the Seattle Seahawks from Paul Allen’s estate for ~$6B or more. Super Bowl winning NFL teams in two of your rivals’ backyard are very Macro Scarce Assets indeed. The man is worth $211.6 billion. He lost $17.8B yesterday alone, or three Seattle Seahawks. Excited to see Zuck talk sports with Theo Von.

But Zuck is spending on some pretty civilizationally useful stuff, too.

On Wednesday, Axios reported that he and his wife Priscilla Chan were committing $500 million out of Biohub, their bio non-profit whose mission is “to cure or prevent all disease,” for its Virtual Biology Initiative. More concretely, the money will go to more compute and more data, which is where money goes these days.

Alex Rives, Biohub’s chief, said they hope to amass a dataset an order of magnitude larger than the current one billion cell datasets, as a step on the “path to building accurate predictive models of the cell.” $400M will go to Biohub’s own work generating massive multi-modal proteomic, genomic, transcriptomic, cellular, and tissue-level datasets, and towards developing the next generation of imaging tools (cryo-ET, advanced microscopy) needed to capture them. The remaining $100M will fund external grants to a who’s who of bio: Allen, Arc, Broad, Sanger, the Human Cell Atlas, the Human Protein Atlas. NVIDIA will be the the technology partner.

To understand why we might want a virtual cell, check out ’s What Are Virtual Cells?

But what do you get for the guy who has record-setting real estate, an NFL team, and a virtual cell?

Space lasers, baby. On Monday, Meta announced that “it signed the first capacity reservation agreement with Overview to receive up to 1 gigawatt of power from the company’s spacecraft.” Overview is Overview Energy, whose founder Marc Berte I spoke with just last week.

The company’s game plan is to put satellites in geosynchronous orbit that beam power to existing solar farms on Earth, so electricity can be delivered wherever it’s most needed without being tied to one location. Unlike Reflect Orbital, which plans to send mirrors up to space to reflect the sunlight back to earth’s solar panels directly, Overview plans to capture the solar energy in space, convert it to laser light, then beam those photons down to existing solar panels, which turn them back into electricity.

I understand that it sounds insane and sci-fi, but I came away from my call with Marc thinking it’s much more near-term than I expected, and Meta’s deal supports that belief.

To get into the nitty gritty, read our friend Rob L'Heureux’s Making Space Lasers Boring in Arena Magazine.

Zuck future so bright he gotta wear shades1.

(3) Rainmaker Proves It Can Make it Rain

Augustus Doricko

I’ve probably covered Rainmaker in the Dose more than any other non-portfolio company, because being able to control the elements for humanity’s benefit is awesome, and because they keep putting out good news with good videos.

This time, CEO Augustus Doricko announced that “Rainmaker is the first company in history to routinely, unambiguously, modify the weather. Last quarter, we produced >143MM gallons of unambiguously man-made precipitation.”

While others have made it rain, Rainmaker is the first to use radar and satellite data to validate that they were actually responsible for the rain, and just how much rain they produced.

Make it rain. Count the drops.

(3b) Skydio Commits $3.5 Billion to U.S. Drone Manufacturing Expansion

Rainmaker makes it rain by sending drones into clouds. Zipline uses drones for delivery. The military uses drones. Law enforcement uses drones. Infrastructure owners use drones. Drones are everywhere, most of them are made in China, and there will be a whole lot more of them in the future than there are now.

To that end, Skydio announced that it plans to invest $3.5B over the next five years to expand American drone manufacturing, including spending over $1 billion with U.S. suppliers. The headline program, SkyForge, is a fifth manufacturing facility five times larger than the existing four (Skydio outgrew four facilities in eight years), with select suppliers invited to co-locate inside it.

We’re gonna need a lot of drones. Skydio is the best American bet. My friend Molly O’Shea went to Skydio HQ to see how they do it.

(4) Colossal Announces the De-Extinction of the Bluebuck Antelope

“More than 200 years ago, humans erased an iconic species.”

“They’re yet another example of a species that has gone extinct by interacting with us.”

“We are excited to announce the de-extinction of… Hippotragus Leucophaeus.”

Colossol Biosciences, of Wooly Mammoth and Direwolf fame, plans to bring back the Bluebuck. Just because we can absolutely means we should.

(5) How American Dads Became the Parents Their Fathers Never Were

and

What’s the point of all of this time-saving and life-extending technology if we can’t use all of that time and life on the important stuff. Luckily, we (dads) seem to be, according to Derek Thompson and Aziz Sunderji, who ran the time-use data on three generations of American dads.

In 1965, the typical married father spent ~30 minutes a day actively engaged with his kids. Today, Millennial dads spend more than 80. That's roughly triple the Boomer baseline and nearly quadruple the Silent Generation. Mothers' time with kids went up, too. The new childcare hours are not coming out of the mom column.

The piece walks through four explanations, none of which fully covers it on its own. The mass entry of women into the workforce is the obvious one but the timing doesn't quite work. Dads also seem to like it. There’s Rameys' "Rug Rat Race," where intensive parenting is partly status competition, partly anxiety about a kid's path through college and the labor market. And as Marc Dunkelman has written, the decline of community and weak ties in America has pushed leisure indoors, which means dads are physically around more, which means they're parenting more.

One that they don’t really talk about but feels right is technology. I got to spend way more of Dev and Maya’s first few years with them because I was able to work from home thanks to Substack and Zoom. While it’s bad that we’re distracted by phones when we’re with our kids, phones also mean that we’re still reachable at the playground.

The essay’s subtitle captures my own experience pretty well: “The new American dad is more present and more exhausted—but also, more satisfied with life.” Also this: “new parents often have short-term loss of brain volume.” Also this:

I wouldn’t trade it for anything in the world.

EXTRA DOSE: Science Breakthroughs (including a 5-in-1 GLP-1 successor), Earth AI, PTJ, Nicholas Thompson

Read more

Subscribe now

Hi friends 👋,

Happy Thursday!

A few weeks ago in Not Boring Capital’s quarterly LP update, I wrote a small essay on scarce assets in a period of abundance. Then Thrive bought the Giants and HOF bought Bugatti and Marc Andreessen pointed out that when one thing becomes abundant, another becomes scarce, and I figured it was a good time to dive a little deeper on what is going to be a supercycle in scarce assets of all shapes and sizes (and a bear market for the easily replicable).

It is also, unofficially, the latest installment in my ongoing plea that you just be different and more you.

Let’s get to it.

Today’s Not Boring is brought to you by… Deel

Hiring globally doesn’t have to be complicated

Hiring globally can unlock growth, but local laws, payroll, and compliance often slow startups down.

Deel’s free guide breaks down what an Employer of Record (EOR) is, how startups use EORs to hire internationally without opening entities, and when it makes sense to use one so you can scale with confidence.

Download Now

Scarce Assets

The playwright S.N. Behrman, in a short study of the Gilded Age English art dealer Joseph Duveen titled “Duveen,” quipped that “Duveen … noticed that Europe had plenty of art and America had plenty of money.”

Duveen made his fortune and his legend balancing that imbalance. “Joseph Duveen sold hundreds of Old Masters, for soaring prices, to American multimillionaires between the early years of the twentieth century and 1939, when he died, at the age of sixty-nine,” wrote Peter Schjeldahl in The New Yorker. He sold a Rembrandt to Carnegie partner Henry Clay Frick in 1906 for $225,000, which was absurd at the time, and the absurdity only grew. The hundreds of millions, in today’s dollars, of Frick’s money that Duveen spent on the Old Masters formed the basis for The Frick Collection.

Not to be outdone, and the whole point was not to be outdone, the Pittsburgh banker and US Treasury Secretary Andrew Mellon purchased, through Duveen, Raphael’s Cowper Madonna for $970,000 (~$21 million today) in 1929, and then, in 1936, a collection of 42 Italian Renaissance paintings and sculptures for $21 million, or half a billion today. These became core to the National Gallery of Art.

Duveen was an absolute animal, who practically spied on clients and potential clients in order to understand their psychology, their likes and dislikes, and even their trash. Per The New Yorker, “(A Duveen employee crowed that, during the years that Mellon was Treasury Secretary, ‘the contents of his wastebasket reached the train to New York in the time it took the Secretary to walk home from the office.’)”

Duveen knew the value of understanding what was priceless to someone who could afford anything with a price tag. For the newly rich of the Gilded Age, as for the wealthy of any era, that was status via scarcity, and the thing that was high-status and scarce for American millionaires back then was connection to the heritage of the old country.

Humans are funny, and the Robber Barons’ clans had the money to be funnier than most. Alva Vanderbilt, the wife of Cornelius Vanderbilt’s grandson William, was the funniest of them all.

Despite the Vanderbilts’ gargantuan railroad and shipping fortune – the Commodore had more money than the US Treasury when he died in 1877, something like 5% of all the money in the country per The First Tycoon – they lacked social standing in New York. Caroline Schermerhorn Astor and Ward McAllister’s “Four Hundred,” instituted as a response to the meteoric rise in multimillionaires after the Civil War, defined who counted as Society. The Vanderbilts were not on the list.

So Alva did what one does. She built a French chateau on Fifth Avenue that dwarfed the street’s older townhouses, and threw the party of the century: the 1883 Vanderbilt Costume Ball. The point was to get on Astor’s list, so Alva supposedly withheld an invitation from Carrie Astor because Mrs. Astor had never called on the Vanderbilt home. To get her daughter invited, Mrs. Astor had to leave her visiting card at 660 Fifth Avenue, which meant officially recognizing the Vanderbilts. The Astors’ invitation arrived the next day.

The party cost $250,000, or $6 million in today’s dollars, but as The Museum of the City of New York writes, “as of March 27, 1883 the Vanderbilts were at the top of a new New York society that was not just limited to 400 people.”

New York City, evidently, wasn’t enough for Alva, nor for many of America’s multimillionaire class. In 1895, Alva married her daughter, Consuela Vanderbilt, to the Duke of Marlborough. Her money, his title.

By 1915, Titled Americans reported that there were 454 “Dollar Princesses,” or American heiresses who had married into European aristocracy.

Titles were scarce. They weren’t making more of them. And therefore, they were desirable to a rising class of people who could buy almost anything.

As Duveen might have noted, Europe had plenty of titles and America had plenty of money.

Not so fast, says Duveen biographer Meryle Secrest, who thought Behrman’s line was overly simplistic. Whether art or titles, she writes, “Nothing would have persuaded titled Europeans, and particularly the British landed aristocracy, to part with their family heirlooms had there not been a catastrophic change of fortune in the final years of the nineteenth century.”

Why were the Europeans suddenly much poorer than they were classy? What was the catastrophe?

“Cheap food from the United States and elsewhere had wrecked the profits of European agriculture.”

The Europeans were done in by abundance, so they had to sell their Scarce Assets, which became more valuable, thanks to the wealth generated by that very same abundance. Pip pip.

Humans fucking love scarce things. Always have. The more we win from abundance, the more we want to roll the winnings into scarce things. Scarce is special.

If we are entering a period of untold abundance, expect a roaring bull market in Scarce Assets.

Macro Scarce Assets

Which is what we are seeing.

Last Friday, Josh Kushner announced Thrive Eternal, the firm’s permanent capital holding company that will concentrate in a small handful of “Iconic franchises and cultural institutions rooted in tradition, identity, and shared experience,” starting with the San Francisco Giants.

Later that same day, HOF Capital announced that it’s leading a consortium to acquire Porsche’s stakes in Bugatti Rimac and Rimac Group.

They are not making any more 143-year-old baseball teams in the technology capital of the world, nor are they making any more 117-year-old French hypercar houses.

There is an obvious story here that, despite its obviousness, is worth spelling out.

As wealth grows and concentrates, demand for a limited pool of Scarce Assets dramatically outstrips supply.

Since the turn of the millennium, global GDP has more than doubled from $78.69T to $174.28T (in 2021 Dollars), while the number of San Francisco MLB teams and Bugatti-makers has stayed flat.

GDP growth undersells the situation, though, because the composition of the growth is more relevant to our conversation. If the $100T increase in GDP was split evenly across the world’s 8.3 billion people, each person’s extra $12,048 wouldn’t mean squat for the price of the San Francisco Giants. People would be able to spend less of their money on food and more on clothes or vacations, but they wouldn’t be able to buy baseball teams.

More relevant is the fact that the wealth of the world’s Top 100 wealthiest people has grown 10x in nominal terms, from $895 billion in 2000 to $7.2 trillion today. Adjusted for inflation, the world’s richest 100 people are more than 4x wealthier today than they were a quarter century ago, and their wealth has grown twice as fast as the global economy.

Bill Gates was the world’s richest person at the turn of the millennium, with $60 billion. Today, Elon Musk and Larry Page have more money than all of 2000’s billionaires combined. The amount of money that Cambricon CEO Chen Tianshi made yesterday would rank him 60th on the 2000 list. You probably haven’t even heard of Cambricon.

Plus, more of the world’s savings sit inside of professionally managed pools of capital whose mandate is to preserve purchasing power, compound over long periods, and find assets that can’t easily be printed, copied, or competed away. Pension assets have more than tripled since the early 2000s, reaching roughly $70 trillion by the end of 2024. Private markets, a rounding error at roughly $600 billion to under $1 trillion in 2000, are now a $13 trillion to $15 trillion asset class. Sovereign wealth funds, which held roughly $1 trillion in 2000, now control something like $12 trillion to $15 trillion.

So, a much higher and more concentrated numerator (more cash) chasing a ~flat denominator (Scarce Assets).

Compounding the issue, these assets are often taken off the market altogether, shrinking the pool. Frick’s art is in The Frick Collection. Mellon’s is in The National Gallery. Thrive named its new vehicle Eternal.

So, a much higher and more concentrated numerator (more cash) chasing a decreasing denominator (Scarce Assets).

And boom goes the dynamite.

This dynamic not new. All of that concentrated capital has been putting upward pressure on major sports franchise prices for a while. Check out this NBA team value data visualization by Koba Khitalishvili.

It’s interesting to note that even within this small pool, the team that has become the most valuable ($11B) and whose value has increased the most (2,344%) is the one in San Francisco, where top-end wealth has increased the most over the time period.

Per Sportico, the total value of NFL teams grew from $190B to $228B between 2024 and 2025, and perhaps unsurprisingly, the top 10 teams’ value grew 24.3% compared to 17.8% for the bottom 22. There are levels to this, scarcity within scarcity.

I could go on. I will go on. People love wealth porn.

Last year, Vlad Doronin set the Miami-Dade record with the $120 million sale of his Star Island home, pricemogging Ken Griffin’s $107 million 2022 purchase.

That record didn’t last long. In March, Meta CEO Mark Zuckerberg purchased 7 Indian Creek, on the even more exclusive Indian Creek Island, nicknamed “Billionaires’ Bunker,” for $170 million.

Coincidentally, Zuck purchased the home from L.A.-based cosmetic surgeon Dr. Aaron Rollins. I say coincidentally, because in a previous draft of this essay, I wrote “thanks to Instagram-face, full lips and smooth skin aren’t what they used to be.” It is only fitting that someone making previously scarce “beauty” abundant would put the fruits of that commoditization to work in Scarce Assets.

Is $170 million a good price for the neighborhood? How does it comp on a per square foot basis? Who gives a shit? You are asking the wrong questions. $170 million is like one AI researcher. It is less than 0.1% of Zuck’s wealth. It is about one-third of what Zuck and Priscilla committed to create better AI simulations of the human body yesterday, while I was in the middle of writing this paragraph. There is only one Indian Creek Island, and it has only so many lots, especially when you consider how many Jeff Bezos has taken off the market (three).

Take the money from The Everything Store (abundance). Roll it into Indian Creek real estate (scarcity).

And then, of course, there is art. Once used to launder class from Europe to nouveau-riche America, it is now playing the same role in the Gulf.

In November 2025, Gustav Klimt’s Portrait of Elisabeth Lederer sold at Sotheby’s for $236.4 million.

It was the highest price ever paid for modern art at auction, and the second-highest price paid at auction for any art ever after the November 2017 Christie’s sale of Leonardo da Vinci’s Salvator Mundi for $450.3 million.

While both buyers were anonymous and neither has been confirmed, the strong rumor is that Saudi Crown Prince Mohammed bin Salman (MBS) bought Salvator Mundi, and those in the know believe either MBS or Abu Dhabi purchased the Klimt. If the latter, it would reportedly be to anchor the collection at the Frank Gehry-designed Guggenheim Abu Dhabi.

Not just any country gets a Gehry-designed Guggenheim, you know.

Is Portrait of Elisabeth Lederer “worth” $236.4 million? You are missing the point. Who gives a shit? Is it even one of Klimt’s top 10 works? Artnet didn’t think so in January 2025, before the piece came to auction, but who cares. Klimts don’t go up for auction every day, and exchanging abundant dollars for scarce Klimts is a trade you do every day and twice on Sunday.

These things are like some hyperVeblen Goods - not only are they more desirable the higher the price, the high price is the entire point.

I am having a lot of fun writing this and I would love to keep going, but I have the rest of the essay to get to.

Everything we’ve covered so far fits into a bucket I’d call Macro Scarce Assets.

What I mean by that is that if the top-end of the wealth distribution keeps getting richer, the prices of these assets will keep going up. There is global competition for them. The buyers discussed in this section include multi-billionaires from the United States of America, Russia, and the Middle East.

If we want to include Asia, we might throw the $600 million wedding Indian industrialist Mukesh Ambani threw for his son Anant in 2024, the closest thing to the Vanderbilt Costume Ball this decade.

If you have the money and access, and you believe that we are embarking on the Singularity, I recommend that you buy as many Klimts, NFL Franchises, and Yellowstone Club homes as you can get your hands on.

I don’t though, sadly, so I’ve been spending a lot of time thinking about how a little guy like me might participate in the abundance → scarcity trade, and I’ve come to the conclusion that it’s everywhere.

Positional Scarcity

Alex Danco, in one of my favorite essays ever, wrote: “In conditions of abundance, relative position matters a great deal.”

He then grouped the different flavors of positional scarcity into categories:

What we’ve called Macro Scarce Assets mostly fit in the Prestige piece of the chart, including where it overlaps with Access to become Proximity and Curation to become Legitimacy.

Again, the useful way to think about Macro Scarce Assets is that as long as the rich get richer, they will get more valuable. “Relative position” here refers to the relative position of the asset owners versus each other based on the Scarce Assets they are able to accumulate.

But there is a different kind of scarcity that exists in relation to other assets, and moves as those assets’ relative abundance and scarcity changes. These are more specific. They are Micro Scarce Assets.

Like, when the Printing Press makes printed text cheap, handwritten text gets more valuable.

Federico da Montefeltro

In the 15th century, long before the Europeans were reduced to selling either their titles or their scarce things, Federico da Montefeltro, the Duke of Urbino, built one of the world’s greatest libraries.

Close readers will appreciate the century: it’s the same one in which Johannes Gutenberg invented the printing press. If you were a Duke looking to fill a library, I mean, what a gift!

Except that Federico da Montefeltro, the Duke of Urbino, refused to taint his shelves with even a single printed book.

His bookseller Vespasiano da Bisticci wrote, “In this library all the books as superlatively good, and written with the pen, and had there been one printed volume it would have been ashamed in such company. They were beautifully illuminated and written on parchment.”

Printed text became abundant and cheap, which made handwritten works more scarce and valuable.

The Duke attempted to do two things at once: first, as da Bisticci writes, “to do what no one had done for a thousand years or more; that is, to create the finest library since ancient times,” and second, to do it all by hand. “It is now fourteen or more years ago since he began the library,” fawned da Bisticci, “and he always employed, in Urbino, in Florence and in other places, thirty or forty scribes in his service.”

After his death, da Montefeltro’s successor dukes, the della Rovere, continued the handwritten-only tradition. “They continued to collect codices, even to have printed books copied by hand (a Borgesian touch), since only codices could enter this hallowed hall, and by the time the Library went to Rome there were 1,760 volumes.” This detail comes from Roderick Conway Morris’ coverage of the 2007 Federico da Montefeltro and His Library exhibition at the … Morgan Library.

The Morgan Library, which itself holds a Bezosian trio of Gutenberg Bibles, was built on the same impulse to collect the irreproducible in a time of abundance that motivated the Duke, and the same desire to import heritage that drove Mellon, Frick, Vanderbilt, and the other Gilded Age Industrialists. Human nature is remarkably consistent.

The da Montefeltro Library was doing something different than the Morgan Library or the Frick Collection. It was specific. When printing made text abundant and cheap, it made handwritten codices scarce and valuable.

Micro Scarce Assets

Marc Andreessen spent part of his weekend doing what he does best: memeing an idea into the mainstream through repetition. In this case, “When something becomes abundant and cheap, another thing becomes scarce and valuable.”

Macro Scarce Assets are structurally scarce: their supply is fixed or shrinking while global wealth rises. Micro Scarce Assets are relationally scarce: they become valuable because something adjacent becomes abundant. Some things can be both.

If Macro Scarce Assets are like blowing up a balloon, it just keeps getting bigger, then Micro Scarce Assets are like squeezing a balloon, the air has to move from one place to another.

Clayton Christensen nailed the Micro mechanism down most tightly in his Law of Conservation of Attractive Profits. From The Innovator’s Solution via Stratechery: “The law states that when modularity and commoditization cause attractive profits to disappear at one stage in the value chain, the opportunity to earn attractive profits with proprietary products will usually emerge at an adjacent stage.”

When one particular thing becomes abundant and cheap, this other specific thing becomes scarce and valuable.

It is an important lesson to keep in mind, one of those ideas that’s as close to a law of business physics as business gets. This is why Joel Spolsky observed that “Smart companies try to commoditize their products’ complements.” Make the thing next to you more abundant and cheaper, so you become scarcer and more valuable.

As a non-but-aspiring-billionaire, the hunt for potential Micro Scarce Assets is where I spend most of my time. These you can catch before they actually become scarce, when they’re not priced scarce.

This was basically the theme of Power in the Age of Intelligence, the hunt for companies that could use technology to capture a scarce position, expand outward from there, eat the industry, and become Scarce Assets themselves.

Real estate is one of the places this is easiest to see. The supply of land on the Earth is basically fixed, so you’re betting that demand for your particular piece of it will increase, and you can do things to make that happen.

If you’re an individual, you might build a very nice house on your land. If you’re a developer, you might put in shops and restaurants and a walkable Main Street. If you’re Japan’s JR-East, you build trains through and to the land, and then build apartments, hotels, and shops on the land, so that you can physically get more people there and convince them to stay.

If you’re Proto-Town, you buy a bunch of land 30 minutes outside of Austin, make it incredibly fast and easy to build things there, and attract a group of tenants / residents who can also build most of what the town needs, from houses to power.

On Friday, the same day that Thrive announced Eternal and HOF announced Bugatti, Proto-Town had its coming out party with that Core Memory video and a WSJ article. This is Micro Scarcity at work: as capital for hard tech and ideas about what to build become abundant, a place you’re actually allowed to build fast becomes scarce and valuable.

Land near fast-growing cities is structurally scarce, but it is more valuable to a group like Proto-Town than it would be to someone who just wanted to buy and hold, because Proto-Town can add value.

The right Scarce Asset to the right buyer basically has a scarcity-set floor and an abundance-uncapped upside.

Take the Giants purchase.

Assume for a moment that AI researchers like sports, and imagine Thrive letting its portfolio companies use its newly-acquired baseball team to recruit those researchers: owner’s box, batting practice with the team, whatever. All hypothetical. The value of a top-end researcher, as set by Zuck, is somewhere in the hundreds of millions of dollars. How do you price a small edge in recruiting? Or in capital raising? Or, if Thrive is able to beat out a rival firm for a deal on the margin because it owns the Giants, and that deal ends up returning $5 billion, did they basically get paid to own the Giants?

The more uncapped your upside, the more valuable the right Scarce Asset is to you.

If this sounds silly, it was the logic for OpenAI’s acquisition of TBPN for a reported “low hundreds of millions.”

Obviously, OpenAI was not buying TBPN’s $30 million in revenue, because it stopped ads immediately. OpenAI bought a universally-beloved show hosted by two of the most commercially creative people in tech, tech media’s Scarce Asset.

The way Scarce Assets work, I doubt OpenAI had a list of ten tech media platforms with some valuation-math-based number it would pay for each. At 900 million weekly active users, ChatGPT has wider distribution than any media property in the world. What it didn’t have, which TBPN did, was likeability. OpenAI should be willing to pay almost anything for that.

More dumb math: at OpenAI’s $852 billion valuation, a 1% lower chance that the company is regulated out of existence is worth $8.52 billion. A 1% bump in its eventual public market price is worth, give or take, $8.52 billion. If there’s even a 1-3% chance TBPN helps deliver either of those 1% swings, it’s worth low hundreds of millions to OpenAI.

If that sounds silly, consider the Elon Musk Family of Companies.

A few weeks ago, I was walking around Washington Square Park with a friend and we were talking about whether Anduril would suffer the fate of recent tech IPOs like Figma and Navan if and when they go to market. My bet is that it wouldn’t, because it is a Scarce Asset.

Like the asset itself, the equity, is scarce, in a way that most post-IPO tech stocks are not.

There’s this thing I’ve noticed for a while but never put into words or numbers, that startups are incredibly sexy while they’re in the private markets, and then become pretty boring within months of IPO.

This is separate from the idea that the private markets overvalue startups and the public markets look at them clear-eyed and set the rational price. I am talking about an aesthetic thing, the butterflies you get looking at a top startup versus the ick you get looking at the same exact company on the public markets a few months later.

Look at the performance of the US venture-backed non-bio companies that have gone public since 2025. The chart shows their performance versus the Nasdaq Composite since Figma’s IPO day on July 31, 2025. Every single one of them has underperformed the Nasdaq over that period.

And there are a lot of good reasons for that. Maybe these companies were overvalued in the private markets. Maybe they popped too hard at IPO and their prices have floated back to reality. Maybe their earnings lag their narratives. Maybe they’ve gotten caught up in the broader SaaSpocalypse. There are a million potential reasons, but one that I haven’t heard but do believe is that they go from being relatively scarce in the private markets to undifferentiated, as assets, in the public markets.

Like, could you explain what Figure Technologies does that’s different from what Coinbase or Robinhood or eToro does? Given a whole entire universe of stocks you could own, why would you buy Navan in particular? Would you rather own Figma as your bet that software will grow, or would it make more sense to buy the basket of software companies and forget it?

In the private markets, a great startup is sexy. It’s a narrative, a secret, a status symbol, a thing you had to win your way into. Once it goes public, it becomes a ticker sandwiched between thousands of other tickers in your brokerage account, sortable by revenue multiple, gross margin, growth rate, blah blah blah. It will come up as one of 248 of companies in a screener for “software companies with 50%+ margins growing 30%+” or whatever. Even if the company hasn’t changed at all, the asset has.

The asset gets compressed down to the boring stuff. It gets Tickered.

This sounds dumb. I get that. But we are so overwhelmed by the abundance of everything that if something wants our attention, it needs to rip it from us.

I’m not commenting on any of these companies in particular, mind you, just trying to illustrate the point that as public companies with tickers, they feel less differentiated. There are other things you could put in your portfolio that could fill a similar role. There are many ways to bet on software.

If you want to bet on space, though? Well, you’re going to have to buy the SpaceX IPO.

SpaceX the stock is a Scarce Asset because SpaceX the company does something rare and valuable, something that cannot easily be copied (and Elon constantly reminds people of that through his owned platform). The micro-scarce leads to the macro-scarce.

This is why, while those other companies struggle in the public markets, with a combined market cap of ~$125 billion across the twelve of them, SpaceX plans to go public at a market cap between $1.75 trillion and $2 trillion. This, despite the fact that those 12 companies combined to do roughly double SpaceX’s $15 billion in revenue last year.

The scarcity of the business and the scarcity of the asset feed back on each other.

Because the business does something rare, valuable, and defensible, and because Elon Musk knows how to use his platform to make it seem even more rare, valuable, and defensible, investors treat the stock as a Scarce Asset. Because investors treat the stock as a Scarce Asset, the company has a lower cost of capital with which to do things that make it more rare, valuable, and defensible.

And then, because the company is worth more, it can do things that have the potential to make it more valuable in reality, like ~acquiring Cursor for $60 billion. As Kevin Kwok explains, the logic for the deal is tight. It would have been harder to pull off if SpaceX were worth ~10x revenue like the other recently-public companies.

Scarce Assets have outsized value today even relative to the actual economic value their moats provide.

I think Anduril is much closer to a SpaceX than it is to a Navan. To bet on the modernization of defense, you can’t buy the Nasdaq or the DJIA or Boeing. Anduril stock, I argued, will be a Scarce Asset.

After I had that conversation with my friend, Primary’s Jason Shulman tweeted this:

It was cool to see Vertical Integrators called out as its own category, and as I wrote in Part IV of the Vertical Integrator Series, the potential to become scarce from a competitive perspective is why I like them so much: “The biggest advantage may be this: because it’s so difficult, and because the advantages you accumulate by doing a million hard things well form barriers, there will be very little serious competition awaiting on the other side of the Great Filter.”

That also means the winning Vertical Integrators have the potential to be Scarce Assets, meaning that not only will their business benefit from micro-scarcity, but they will be able to more easily attract capital thanks to macro-scarcity.

One of the things that’s interesting about Vertical Integrators is that while they share an overall approach, each one can be a unique and scarce asset. Dandy and Flock, for example, are both Vertical Integrators, but Dandy is attacking the dentistry supply chain and Flock is supplying law enforcement with better data. Each can be a Scarce Asset - the best company in its category, and the best way to bet on that category - in a way that almost no pure software company can.

This is another reason I think more value will accrue to a smaller number of companies that win their category. It’s soft and squishy and irrational, but so are we.

Way back in COVID times, in October 2020, to be precise, when I was sleep deprived with three-week-old Dev, I wrote this piece called Software is Eating the Markets that both holds up and doesn’t. A lot of the specific examples look very COVID in hindsight, but I think the core idea, that retail investors, like angel investors, pay for more than just future cash flows when they buy an asset, holds up.

It may not just be retail investors, either. There is so much demand for certain private companies’ secondaries and so little for others among family offices and institutional allocators that startup shares feel like Veblen Goods: there is more demand the higher the price goes. Check out Setter’s list of “the most sought-after venture-backed companies in the global secondaries market.”

This, I think, is how a market behaves when it views certain companies’ equity as Scarce Assets as opposed to purely financial instruments. If you tell someone at a party that you invested in Anthropic or SpaceX, chances are they will know what you’re talking about and perhaps even think that it’s cool! If you buy them now, even if you aren’t going to make 100x, you’ll get to live through the thrill of an IPO that everyone’s talking about with skin in the game.

People are status-seeking monkeys, Eugene Wei wrote, and still we keep underestimating it. This status-seeking shows up even in things that should be purely spreadsheet-based, like late stage investments, and even in hard-headed industrialists like Henry Clay Frick. We value what is scarce.

As more things that were traditionally scarce become abundant, we’re going to keep seeing this playing out across… everything.

Thanks to GLP-1s, skinny isn’t what it used to be. So what will be scarce when you can take a shot or a pill to look fit? As popular as marathons and fitness competitions have become, I bet they become even more popular. You can’t fake a sub-2 hour marathon.

Thanks to AI, generic knowledge has become free and abundant. Meanwhile, prices for top-end rare physical books have jumped. Per the delightful Rare Book HUB, the cutoff to make the year’s Top 500 sales jumped from $81,250 in 2023 to roughly $120,000 in 2024, while the number of lots selling for more than $1 million rose from 12 to 29. Interestingly, the median price has decreased a bit, while the top got hot.

If we are successful in creating the abundant near-future that has been promised, the competition for Scarce Assets will continue to intensify.

There are some Scarce Assets that only a lot of money can buy, like sports teams and rare art. I bet we will see more tech billionaires, or tech companies, buy F1 teams if and as they become available.

But there are many more to be built than there are to be bought.

Unprecedented abundance will create unprecedented demand for things that cannot be made abundant.

That might mean local things: really unique restaurants, theaters, hotels. It might mean new towns altogether. I have a hunch that drones, EVs, and EVTOLs should expand local frontiers. For an enterprising new settler, it might make sense to start gobbling up cheap, buildable land where it doesn’t currently make much sense, or open up new frontiers beyond the land. You might get it wrong, you probably will. No one said scarce was easy.

It might mean personality-driven media, like TBPN, but it certainly won’t mean something that looks like TBPN. People will always want to outdo other people, but we will also like and trust each other more than we like and trust the machines.

You know what, I’m going to stop listing things.

The scarce and valuable thing you create will not come from someone else’s list.

This is the temptation, and the trap. The trap of this era of abundance will be believing that you can do great things easily, that you can create things other people truly value at the press of a button. You can’t.

All of this sounds so obvious when you write it out that I almost didn’t want to hit publish. More money buy scarce things dur dur. When all same do different har har.

But this is how almost no one is acting. People get the abundance button and they start mashing it, like those rats in that experiment, believing that the thing that is now easy for them is only easy for them, that there are no second-order effects, that the world is a static place. That maybe, hey, just for me, greatness will be easy.

Whoops. The millisecond something becomes easy, the value shoots elsewhere. Scarce is hard and hard is scarce, forever and ever, amen.

That’s all for today. We’ll be back in your inbox tomorrow with another Weekly Dose.

Thanks for reading,

Packy

Hi friends 👋,

Happy Friday and welcome to our 190th Weekly Dose of Optimism!

Busy week: cures for cancer, autonomous everything, Sixers actually stole a win off the Celtics, lots of Extra Doses. No time to waste…

Let’s get to it.

Today’s Not Boring is brought to you by… Silicon Valley Bank

Climate Tech is in a weird spot in 2026. The problems are getting bigger, but the money is getting tighter. 

SVB’s latest Future of Climate Tech report shows a market that’s easy to misunderstand right now:

• $29B was invested in Climate Tech last year, one of the biggest years ever… however, most of that funding went to a only a few companies

• Government support is pulling back, making things slower and harder

• 52% of companies are cutting burn

• The conversation is getting quieter. Mentions of climate on public company earnings calls are down 70%.

The market is cutting hype, forcing discipline, and rewarding the companies that actually build. Which is uncomfortable in the short-term, but probably exactly what the category needs long-term.

If you want a clear look at where Climate Tech actually stands in 2026, read SVB’s Future of Climate Tech report.

Get your free copy

(1) Pancreatic cancer mRNA vaccine shows lasting results in early trial

Kaitlin Sullivan, Marina Kopf and Anne Thompson for NBC News

Pancreatic cancer is often a death sentence. It kills about 87% of the people it’s diagnosed in within five years. Worse, it’s sometimes called the “silent killer” because symptoms don’t show up until very late, and by the time they do, only about 1 in 10 patients has a tumor that’s still operable. For the rest, there are no good options. And for a long time, oncologists thought the disease was biologically incapable of mounting a meaningful immune response at all. Pancreatic cancer was supposed to be the cancer that immunotherapy couldn’t touch.

This week at the AACR annual meeting in San Diego, Dr. Vinod Balachandran of Memorial Sloan Kettering presented six-year follow-up data from a Phase 1 trial of a personalized mRNA vaccine for pancreatic cancer that strongly suggests otherwise.

The trial enrolled 16 patients with early-stage, operable pancreatic cancer. After surgery, each got a custom-built mRNA vaccine (autogene cevumeran, developed by BioNTech and Genentech) alongside the checkpoint inhibitor atezolizumab and standard chemotherapy. Each patient’s tumor was shipped to BioNTech in Germany, sequenced to identify up to 20 mutations most likely to look foreign to the immune system, and then turned into a one-of-one vaccine that was shipped back to New York and infused into the patient.

Eight of the sixteen patients mounted a T-cell response to the vaccine. Six years out, seven of those eight responders are still alive (87.5%), versus two of the eight non-responders (25%). One of those seven — Donna Gustafson, who was the first person in the world to receive one of these vaccines in February 2020 — just hiked Mt. Etna for her 50th wedding anniversary. The immunologic data are equally encouraging: 85% of the T-cell clones primed by the vaccine persisted into memory phase, and the memory T cells were still functional even after chemotherapy, producing cytokines and attacking cancer cells on re-exposure to the original neoantigens.

Of course, there are caveats. This was a 16-person Phase 1. There was no randomization. Half the patients didn’t respond at all, and we don’t know why yet. Operable pancreatic cancer is already the most survivable version of the disease, and a subgroup of patients always lives longer than expected. Phase 2, now running globally under Genentech and BioNTech, will tell us a lot more.

With those out of the way, this is awesome, at the very least for these seven people and their families, and hopefully for a lot more. The belief for years has been that pancreatic tumors were an immune desert because it has too few mutations to flag and too dense a stromal wall to penetrate. This trial suggests the wall can be climbed, if you give the immune system the right map: a personalized neoantigen set, manufactured in weeks, and delivered as mRNA.

As Elliot wrote in Going Founder Mode on Cancer, personalized cancer therapeutics hold a ton of promise, and they’re both expensive and complicated for wide-scale usage today. That said, these are exactly the kind of thing that will get radically cheaper and faster to produce as the underlying biotech stack matures.

If this holds up at scale, it’s a twofer for our anti-cancer canon. One: pancreatic cancer is not immunologically untouchable. Two: the personalized mRNA playbook that Moderna/Merck have been running in melanoma is generalizable, and generalizing fast.

As we like to say here at the Dose: get fucked, cancer.

(2) Cursor and SpaceX: In Search of a Complete Loop

Kevin Kwok

Speaking of two-fers… SpaceX basically purchasing a call option on Cursor for $10 billion is the kind of story we would have covered in the Dose anyway, and we get to throw in the first Kevin Kwok essay in nearly a year for free.

On Wednesday, SpaceX and Cursor announced that the newly combined SpaceX / xAI was basically buying a $10 billion option to buy Cursor later this year for $60 billion. SpaceX will most likely buy Cursor after its planned IPO for $60 billion. If it doesn’t, it will pay Cursor a $10 billion breakup fee / fee for the work they did together this year.

It sounds a little confusing, which is why I was so happy to see Kevin explain the deal, both from a deal mechanics perspective and the strategic perspective.

On the deal mechanics side, he expects to see more deals like this.

On the strategy side, the deal makes a lot of sense. Cursor has a great coding product and model, but not enough compute to compete with Anthropic and OpenAI. SpaceXAI is really good at building out compute, but pretty bad at making coding models. Put the two together, and you might have something that can compete with the leading labs in a use case that they all agree is incredibly important to win.

In short, the world’s greatest rocket company bought a product that can write code as it prepares to colonize the Moon and then Mars, obviously.

(3) Fervo Energy Files S-1 for IPO

M&A. IPOs. Oh my. We are back.

Last week, we led the Dose with Quaise breaking ground on the world’s first commercial superhot geothermal plant. Superhot is a drilling problem, and Quaise is looking to solve it.

This week, the company doing the most to solve geothermal’s economics problem at commercial scale filed to go public. On Friday, Tim Latimer and Jack Norbeck’s Fervo Energy filed an S-1 with the SEC to list on Nasdaq under the ticker FRVO, with J.P. Morgan, BofA, RBC, and Barclays leading.

It’s great to see a new clean power generation source hit the public company milestone, with a real, traditional IPO and all, because it will hopefully mean more money for the space. But it’s also fun because it gives us a better look into what Fervo is up to.

Fervo is operating or developing 3.65 gigawatts of geothermal capacity, a number that, if fully built, would nearly double the total installed geothermal capacity in the United States. Its flagship Cape Station project in Beaver County, Utah, breaks ground to first power later this year; 500 MW is under construction at the site and Fervo has permits in hand for another 1.5 GW on the same piece of land. There’s also a 150 MW site in Nevada under a Google + NV Energy deal, targeting 2030.

The economics are even more interesting, because they give us a firmer sense of where geothermal sits today and where it’s headed. Fervo says Cape Station will deliver carbon-free 24/7 power at $7,000 per kilowatt of installed capacity, which is in the same ballpark as advanced nuclear. The company’s goal is to cut that to $3,000/kW, which would beat natural gas on an unsubsidized basis for always-on power. That would be a massive unlock.

Fervo is able to pull this off by bringing well-worn horizontal drilling and completion techniques over from shale (Latimer was a drilling engineer at BHP before founding Fervo in 2017) to geothermal.

To which we say, drill baby drill and let’s get this frackin’ thing public.

(4) Humble unstealths with a cabless autonomous electric truck & $24M

Ana-Maria Stanciuc for TheNextWeb

On Tuesday, a San Francisco startup called Humble came out of stealth with $24 million in seed funding and a design that is so obvious once you’ve seen it that it makes you wonder why no one had done this before. Basically, they got rid of the cab where human drivers sit, because these trucks won’t have human drivers.

Every existing autonomous trucking company, including Aurora, Kodiak, Waabi, and Einride, retrofits driverless tech onto a tractor-trailer architecture designed around a human sitting in front of a steering wheel. As founder Eyal Cohen put it, “Trucks were never designed to be autonomous.”

Humble built one that was. Remove the cab, and you get 360-degree sensor coverage, a significantly lighter vehicle, and a geometry you can optimize for intermodal containers. The Humble Hauler is built for 40- and 53-foot shipping containers and goes dock-to-dock, unloading at the destination rather than handing off to a human driver at a yard near city limits like Aurora does. Eclipse led the seed with participation from Energy Impact Partners.

Cohen, Humble’s CEO, has been on this problem for a while, and on hardware even longer. He previously worked at Apple, Uber ATG, and Waabi, and co-founded Spark AI (acquired by John Deere in 2023). That experience helped him get to a prototype in under 6 months.

This is an idea that keeps popping up as we talk about autonomy: when you design around the actual constraint instead of the inherited one, you end up with a different machine. Kalanick’s Atoms is explicitly anti-humanoid because most jobs want wheels, not legs. Tesla’s Cybercab has no steering wheel, because they won’t have human drivers. Humble is making an analogous bet with trucks.

Better, faster, cheaper. Ship it.

(5) Medra Launches Robotic Bio Lab

for & Michelle Lee on X

Here in the Dose, we’ve written a lot about the software side of the AI-for-bio boom: IsoDDE can suggest drug candidates against previously undruggable pockets, Evo 2 can design entirely new genomes, Boltz and AlphaFold have collapsed structure prediction, and Chai is cranking out antibody candidates at a pace that would have sounded insane five years ago. It’s all fantastic and futuristic.

Having said that… a designed molecule is not a validated molecule. Every one of those candidates still has to be synthesized and tested by a physical lab staffed by physical humans who have to sleep and take weekends. Or, I guess that’s an assumption, certainly these candidates need to be synthesized and tested in a physical lab, but…

This morning, Michelle Lee’s company, Medra, formally opened the doors on a 38,000 square foot warehouse in San Francisco where roughly a hundred robotic arms are running biology experiments simultaneously, 24/7, linked by a small courier robot that ferries pipette tips and sealed plates between stations like an extremely focused junior scientist. A pair of scientists could run maybe a dozen antibody-binding experiments in a day. Medra’s arms run hundreds at a time, don’t clock out, and get better at it every week. We love a good robot arm.

Lab automation has been overpromised for two decades. What’s different about Medra is that it doesn’t require the lab to buy new robot-native instruments. Only about 5% of bench equipment in biology has APIs that legacy automation can plug into; the other 95% (centrifuges you open by hand, pipettes you tilt and time, etc…) was built for humans. Medra’s robots use computer vision and manipulation models to operate the same instruments a human would. Lee says the combination can push the share of bio-tech tasks that can be automated from 5% to 75%.

There’s a second layer on top of the hardware that is maybe even more interesting. Every arm and bench is instrumented with cameras and nine kinds of sensors, so the system logs the exact pipette angle, the exact depth of insertion, the exact timing between reagent additions. The AI scientist on top reads the results, proposes protocol changes, and rewrites the protocol itself. Lee describes one customer whose antibodies weren’t binding at all (the readout came back at zero); the AI scientist narrowed the cause to two hypotheses, proposed adding a vortexing step, and watched binding jump to over 70%.

Lee’s frame for what she’s building is TSMC for biology. TSMC is what makes it possible for chip designers to exist without owning a fab. Medra wants to be what makes it possible for drug discovery companies to run experiments without owning a wet lab.

Just yesterday, in The Great Blue Frontier, Will and I wrote that AI for bio is data-limited, and suggested the 91% of sea creatures we’ve yet to discover as one source. Madra is another. Lee believes that with more throughput, we “cure cancer, Alzheimer’s, infectious disease.” This, she thinks, is also how we keep American bio competitive with China by speeding ourselves up.

All sounds great. Get to work, robots.

EXTRA DOSE: Science Breakthroughs, Oceans, Flipbook, Tokyo & NYC Subways, and a Boots Manifesto

Read more

Subscribe now

Hi friends 👋 ,

Growing up, I sailed a little sailboat called a Sunfish. I was OK, not great, which is a shame, because had I been a little more competitive, I would have raced against the next town over, whose team included a young Taylor Swift.

When I turned 16, I became a beach lifeguard, which is pretty much the greatest job ever for a 16-year-old, and I held onto it for the next four summers. I wasn’t a great swimmer or a great rower, but I was a good enough runner to make up for it so they kept me around. Over those summers, I fell in love with Wes Anderson’s The Life Aquatic with Steve Zissou, which was based, red knit cap and all, on the life of Jacques Cousteau.

All of which is to say, while I spent a lot of time by the ocean, I had a tourist’s relationship with it. I enjoyed it, swam in it, sailed in it, and then went back to the real world, where classes and eventually a real economy awaited.

This, Will O’Brien argues, is how we’ve all handled the ocean until now. We come, we explore or exploit, and we get back to dry land. This, Will thinks, is no longer how we will handle the ocean. With Ulysses, The Ocean Company of which he is co-founder and President, Will plans to help build the infrastructure that would allow us to treat the ocean as a permanent fixture of the economy, and potentially even as a new home for humanity. The way our grandparents and JFK thought we would.

Since meeting Will two and a half years ago, he’s become one of my favorite people in tech to talk to about everything from Irish omnipresence to religion to aliens to vertical integration. I’ve been asking him to put his gift of gab to paper with me for a while, and now that the company has successfully raised $46M from a group of investors led by a16z American Dynamism, he finally had the time to oblige.

Note: not boring capital is not an investor in Ulysses, but I am a huge fan.

In this co-written essay, Will tells a history of the future of the ocean that I’d never heard, before making the case that the ocean is the last great frontier and one of the greatest economic opportunities available to humanity. It’s an adventure, and in the words of Steve Zissou, “Anyone who wants to tag along is more than welcome.”

So throw on the most underrated soundtrack of the 2000s…

And let’s get to it.

Today’s Not Boring is brought to you by… Framer

Launch fast. Design beautifully. Build your company's website on Framer.

Framer helps teams design, build, and launch their marketing sites lightning fast. With the ability to publish hundreds of CMS pages in a single click, operate at a global scale with seamless localization, and even host unified content across multiple domains, teams have never been able to ship faster. Companies like Miro, Bilt, and Perplexity trust Framer to achieve:

• Speed without chaos: ship pages and updates faster without turning the site into a fragile set of one-off hacks.

• Reduced dependency: shift routine brand and marketing work out of product engineering queues.

• Production-grade foundation: Run real marketing systems (CMS, SEO, performance optimization) with governance and collaboration.

👉 Build your company's site on Framer today

The Great Blue Frontier

With Ulysses President Will O’Brien

The Death of the Ocean Dream

At 4 AM on February 17, 1969, Berry Cannon was lowered 610 feet below the surface of the Pacific in a steel personnel transfer capsule to make repairs to the Navy’s “yellow submarine,” Sealab 3. At that depth, normal air becomes toxic to humans and the nitrogen can send you delirious, so Cannon was breathing helium instead. Helium keeps your mind clear, which was helpful for Cannon. He had been awake for twenty straight hours. Unfortunately, helium also strips heat from your body six times faster than air does. This was a problem. That particular morning, the capsule’s heater was broken.

No matter. Cannon, one of history’s greatest maritime frontiersmen, was running on amphetamines and conviction. When the capsule reached the seafloor, he dropped through the hatch into open water and began swimming through pitch darkness toward a leaking underwater habitat. This habitat, the U.S. Navy believed, would be humanity’s first permanent foothold on the ocean floor.

It was a routine fix as far as deep sea habitat fixes go. Except for one fact: Cannon’s rebreather’s CO₂ scrubber canister was empty of Baralyme. He didn’t know that. Nobody did. On a TV monitor topside, Captain George Bond watched Cannon swim gracefully—and then he watched Cannon’s body suddenly jackknife.

“Any time you see rapid motion in a diver,” Bond would say later, “you know he’s in trouble.”

Berry Cannon was dead. And with him died America’s plans to conquer the ocean.

Five months later, Neil Armstrong walked on the Moon, fulfilling one of JFK’s most ambitious promises. It’s been over half a century, and we all know Neil Armstrong’s name.

But nobody remembers the man who died trying to live on the seafloor. Nor do we remember the ocean dream he represented.

I grew up in County Cork, on the south coast of Ireland, and spent every summer at Garrettstown jumping off the pier or the cliffs, bodyboarding and pulling crabs out of rock pools, or heading out in the boat with my dad to catch mackerel. I loved the ocean the way you love something before you understand it. The textures, the cold, the endlessness of it. As a kid, I loved explorers too. Steve Irwin, especially. He wasn’t an ocean guy, but he had that renegade explorer energy, that willingness to just get in there and figure it out. I think I always assumed someone was figuring the ocean out, that there was a Steve Irwin of the Deep. That someone was down there, mapping it, understanding it, and protecting it.

As I got older, I realized nobody was.

So I’ve made the oceans my life’s work. In just the past few months, my office has rotated between Washington D.C., San Francisco, Los Angeles, London, Western Australia, Maine, Virginia, Florida, New Orleans, and San Diego. I would do this for free; I might even pay to do it. But as I’ve been living out my childhood dream, I’ve found that the ocean economy is something like a sunken treasure that’s been waiting on the bottom of the sea for anyone intrepid enough to go grab it.

Today, I’m going to let you in on the secret I’ve discovered down there: Earth’s largest domain and last great frontier is also its grandest economic opportunity.

The Two Frontiers

In the 1960s, America was a country dreaming at full volume. You could taste optimism in the air — and in the sea.

Most of us alive today never learned how important ocean exploration was to the future our grandparents imagined. Americans were fantasizing about ocean exploration in the same way they were about going to the Moon. They were as captivated by The Great Blue Frontier as they were by the Space Race.

At the 1964 New York World’s Fair, GM’s Futurama II ride carried more than 26 million visitors past a vision of the near future. Alongside the lunar base and the Antarctic weather station, guests saw ocean-floor oil rigs, submarine trains hauling minerals to shore, and Hotel Atlantis, a sub-oceanic resort around which vacationers explored in various personal oxygenated craft.

The ocean stood right beside the Moon in the American imagination, and it promised just as much: food and living space for a rapidly growing country, minerals and energy for a booming industry, and the upper hand in the Cold War.

President John F. Kennedy saw the problem clearly. “We know less of the oceans at our feet, where we came from, than we do of the sky above our heads,” he told the National Academy of Sciences in 1963.

His administration had been trying to close that gap from the start. In a March 1961 letter to Congress, he called for “a national effort in oceanography,” warning that “knowledge of the oceans is more than a matter of curiosity — our very survival may hinge upon it.” Kennedy’s FY1962 budget request nearly doubled federal oceanography spending, funded ten new research vessels, and expanded shore facilities fivefold. Two years later, he sent Congress a ten-year, $2 billion plan called Oceanography: Science for Survival, and directed the Navy to begin SEALAB — an underwater habitat program explicitly conceived as the ocean floor’s answer to the Space Race.

President Kennedy placed the ocean alongside space as the twin frontiers of American ambition and coequal national priorities, and his ambition was matched by action.

While Kennedy persuaded Americans about the ocean’s importance as a governmental priority, French naval officer, explorer, and filmmaker Jacques Cousteau charmed them in their living rooms.

Cousteau had co-invented the modern scuba regulator and was one of the most famous people in the world. He built Conshelf, a series of underwater habitats where teams of divers lived for weeks at a time on the seabed. Conshelf II, built in the Red Sea in 1963, was essentially an underwater village: a starfish-shaped structure at a depth of 33 feet with bedrooms, a kitchen, hot showers, and a television.

A parrot named Claude served as the carbon dioxide detector. If Claude fell off his perch, the air was bad. You couldn’t whistle because of the helium. Matches wouldn’t light, although the crew still managed to light cigarettes. Sparkling wine went flat under pressure. Fried food was forbidden because greasy fumes couldn’t be scrubbed from the air.

The documentary about this village, World Without Sun, won the Academy Award. The Undersea World of Jacques Cousteau ran on prime-time American television from 1966 to 1976. The ocean was mainstream culture, Academy Award-winning cinema, and prime-time television. It had become a fixture of the popular imagination.

Simultaneously, the U.S. Navy was building permanent undersea infrastructure with the same seriousness that the Army had when it built forts across the American West. Its program, SEALAB, consisted of three progressively deeper underwater habitats that were designed to prove that humans could live and work on the ocean floor for extended periods. We were going to settle the ocean. On SEALAB II, astronaut Scott Carpenter, one of the original Mercury Seven, spent 30 consecutive days underwater, becoming the only astronaut-aquanaut in history.

By the late 1960s, more than 60 underwater habitats dotted the world’s seabeds: Hydrolab, Helgoland, Tektite, Aquabulle, Hippocampe, dozens more.

Kids wanted to be aquanauts the way they wanted to be astronauts. The ocean was a national obsession on par with the Moon.

Then, in roughly five years, it all died.

SEALAB was suspended immediately after Cannon’s death. The Navy, which could invest in the future during peacetime, was pulled into the Vietnam War. Even Jacques Cousteau, the man who had done more than anyone alive to prove humans could live underwater, the man who had unmatched cultural authority on anything to do with the ocean, pivoted from exploration to activism. He founded the Cousteau Society in 1973 to protect and preserve the oceans instead of exploring and settling them.

By the 1992 Rio Earth Summit, he was dubbed “Captain Planet.” America, in some sense, always followed Cousteau’s lead on the ocean. When he explored, we wanted to explore. When he said “protect,” oceanic policy and funding priorities fell in line. The regulations that followed, including the Marine Mammal Protection Act, UNCLOS, the IWC whaling moratorium, and the High Seas Treaty, were individually important protections, responding to real ecological crises that demanded our attention. But their cumulative effect, compounded by Cousteau’s cultural reframing, led to a collective shift in how we considered the ocean as a domain to build in. Within a generation, any proposal for persistent ocean activity faced a default presumption of harm.

In the 1960s, America set its sights on two frontiers. By the 2010s, we had all but stagnated on those hopes. We landed on the Moon, went back a few times, and then decided to keep our eyes on the ground. We gave up on settling the ocean altogether.

But we are going back to the Moon, this time to settle it. Just this month, the Artemis II astronauts orbited the Moon for the first time since 1972 in an important step towards settlement. NASA Administrator Jared Isaacman has laid out a plan to begin building a permanent base in 2028, a mission made possible by progress in the commercial space sector over the past two decades. We are going to be an interplanetary species! The Moon should be a state, experts are saying.

It’s time that we settle the ocean too.

This forgotten frontier is no less economically or geopolitically important than space; in fact, it is both larger and more urgently strategic. Through our forgetfulness, the ocean has become a wild and lawless domain whose potential to benefit humanity lies dormant, one which is being abused by those who care nothing for protection or preservation.

What Happens to a Frontier Forgotten

In 2015, a wooden fishing boat washed ashore on the coast of Japan carrying the skeletal remains of its crew. Then another arrived. And another. Over the next several years, hundreds of these “ghost boats” drifted onto Japanese beaches, small wooden vessels with many carrying only bones and tattered North Korean flags.

Investigators eventually pieced together the story, which turned out to be more tragic than supernatural. 900 Chinese distant-water fishing vessels had moved into North Korean waters in violation of UN sanctions (sanctions that China itself had signed and ignored), strip-fishing the stocks and pushing local fishermen further and further offshore in boats not built to take on the open ocean. Those fishermen starved at sea, and their boats drifted east until Japan’s coastline caught them. Nobody stopped the Chinese fleet, and nobody rescued the fishermen, because there was nobody out there to stop or rescue.

A 2024 study in Nature found that three out of four industrial fishing vessels are invisible to public tracking, which is wild in a modern society that tracks everything.

Consider the pandemonium that broke out when MH370 went missing, because planes, unlike fishing vessels, don’t just go missing. Last weekend, I spoke with a man who was in a terrible car accident on a Canadian highway in 1994; his right leg and hip were crushed, but he lived because a helicopter arrived to whisk him to the hospital within five minutes. In the air and on land, we track our people and our critical assets with stunning accuracy.

But the ocean doesn’t operate like that, in large part because we don’t have the infrastructure to make monitoring possible. So we’ve kind of given up. Two-thirds of the ocean falls outside any country’s jurisdiction, and there is no police force or coast guard with global reach. There is a reason that so many ocean movies involve protagonists lost at sea, hoping that someone chances upon them.

As a result, we often find boats doing bad things, too late. Fleets of hundreds of Chinese vessels have been caught operating illegally off the coast of Ecuador, right at the edge of the Galápagos Marine Reserve, one of the most ecologically sensitive places on Earth. In 2017, Ecuador intercepted a single vessel, the Fu Yuan Lu Leng 999, which was carrying 6,000 sharks, many of them endangered species, fished from Galápagos waters. Workers on distant-water fishing vessels are held for years, passports confiscated, in conditions that meet every definition of slavery.

There is no way of knowing how many illegal vessels go undetected, because so much of the ocean remains unmonitored, and we have less of a clue the further beneath the surface we go. While humanity has mapped 100% of the surface of Mars, a planet with no known life 140 million miles and seven months (if you time it just right) away, we have mapped only 27% of the ocean floor with modern sonar, and much of that is coarse, low resolution.

“If you just look at the size of the ocean,” Bob Lazar (yes, that Bob Lazar) recently told Jesse Michels, “you can hide an entire civilization down there. Especially if they’re immune to the effect of the ocean. You just gotta be deep. We’ll never find ‘em.”

While we can neither confirm nor deny the existence of subsea civilizations (except for maybe Atlantis), the fact is there is certainly enough space and little enough visibility to get away with it. The ocean is unfathomably large, and we haven’t even begun to fathom what’s happening all those fathoms below.

This, as President Kennedy warned in 1963, is more than a matter of curiosity. Our survival, or at least our flourishing, may hinge upon it.

Sixty-three years later, we have failed to heed his words, and to capitalize on the opportunity. The ocean covers 70% of our planet, carries 99% of our internet traffic and 80% of our trade. It contains more critical minerals than all known land reserves. And we still know critically little about any of it.

A full 91% of ocean species remain unknown to science. When researchers sequence DNA from deep-sea sediments, they can’t match it to any known organism by species or taxonomic group. The largest library of biological information on earth is functionally unread. Chances are, that library contains a well-stocked pharmacy.

Marine organisms have already given us Ziconotide (a painkiller 1,000x more potent than morphine, from cone snail venom) and Trabectedin (a cancer drug, from a sea squirt). The alien-looking jellyfish Aequorea victoria provided researchers with the green fluorescent protein (GFP) that won them the 2008 Nobel Prize in Chemistry, which now “enables scientists to track, amongst other things, how cancer tumours form new blood vessels, how Alzheimer’s disease kills brain neurons and how HIV infected cells produce new viruses.”

Presumably, there is more where that came from inside the other 91% of sea creatures we have yet to discover. If AI for bio is data-limited, we have a library full of it in the ocean.

There is much for a knowledge-loving people to discover down below, so… let’s just go find out what else is down there. Let’s send people and instruments and start intervening.

Alas, there is nothing to intervene with, because we never built anything that could intervene, and little more to listen with, because we let our ears rot.

Back in the 1950s, back when the nation had aquatic ambitions and a Cold War foe against which to sharpen them, the US Navy built a classified network of undersea hydrophone arrays called the Sound Surveillance System, or SOSUS. By exploiting the deep sound channel, or SOFAR channel, SOSUS could track Soviet submarines across enormous swaths of the Pacific and Atlantic Oceans. Individual listening stations could pick up a submarine across entire ocean basins thousands of miles away.

Turns out, SOSUS could track whales, too. When it was partially declassified in the 1990s, marine biologists realized the Navy had been accidentally collecting the richest dataset on ocean biology ever assembled. They’d captured volcanic eruptions, seismic events, whalesong, species migrations, and, in 1997, the “Bloop,” a mysterious sound that has still not been fully explained…

But the Cold War ended in the ‘90s, and by then, our ocean dreams were already long dead. The US government declassified SOSUS, celebrated its scientific contributions, then cut its funding. Some assets were folded into another program, IUSS, a bunch of the hydrophone arrays were put into standby status, and stations at places like Bermuda, Adak, and Keflavik were shut down.

Today, our greatest effort at ocean observation is a network of 4,000 robotic floats called Argo, the 27-year-old “crown jewel of ocean observing systems.” Each of the 4,000 floats monitors an area larger than Portugal, surfacing once every ten days to transmit a single temperature reading. They are deployed like this…

Or, like this...

The entire program is tax payer-funded. And despite running on just six cents per American per year, those 4,000 floats now generate more subsurface ocean data every month than the entire rest of the observing network combined. Argo is great, but that is much more an indictment of the state of our ocean awareness than a celebration of the system.

Plus, Argo can only observe, just like SOSUS could only listen. If a float detects a chemical anomaly, a biological collapse, or a dangerous trend, there is nothing and no one for miles around to respond, in much the same way a thermometer can tell you that you have a temperature but can’t do a damn thing about it.

The question is, in a time of technological wonder, why the ocean remains dark.

The Ocean is Pre-Industrial

Frontiers are tamed when there’s money to be made from taming them and the technology to tame them is viable.

President Thomas Jefferson sent Lewis and Clark west to wrest the fur trade from the Canadians and establish commercial routes. The Forty-Niners went to find gold. These were risky and potentially highly profitable expeditions, and when they paid off, America built the railroad to connect the West with civilization.

The railroad meant that normal people, not just bold or desperate explorers, could establish their homes, families, and businesses in the West, and it made new industries possible: ranching, mining, commercial agriculture, towns, and cities. The railroad industrialized the Western Front.

Something similar happened in space, a domain accessible only to national governments and a handful of intrepid telecommunications pioneers in search of vast riches before SpaceX collapsed launch costs. Before SpaceX, the biggest satellite constellations were measured in dozens. Today, SpaceX has more than 10,000 Starlinks in orbit; people make fun of Jeff Bezos for flying “only” a couple hundred satellites.

Reusable rockets are the space railroad, and on them fly hundreds of businesses seeking to get to space and stay. SpaceX industrialized space, and the effects of that industrialization are just beginning to be felt.

For all of the exploration and trade we’ve done out there, the ocean has never had this moment. We’ve fished it, laid cables across its floor, drilled rigs into it, plopped wind turbines in it, raced around the globe on it, and shipped goods across it, but nobody has ever cracked the economics of operating within it, broadly and persistently, as a domain. The ocean still runs on an 1800s-era expedition model that puts expensive humans on expensive ships for expensive campaigns, then brings them back to dry land.

There’s a word for a frontier that hasn’t had this moment yet. The ocean is pre-industrial.

To industrialize a frontier means to transform it from a place with industries in it, exploring and opportunistically extracting, to a place built to support industries persistently and economically, at scale.

An industrialized frontier is a platform, not a series of projects. Each new piece of infrastructure makes the next one cheaper and unlocks activities that weren’t possible before. An industrialized frontier compounds.

The railroad industrialized the West and SpaceX is doing the same for space. Nobody has industrialized the ocean yet.

If we want to gain dominion over Earth’s oceans and steward them properly, we need to make it a place where industry can thrive.

Creating more industry in the ocean would decrease crime, increase our understanding of our home planet, accelerate scientific discovery, grow existing maritime industries and enable new ones altogether, unlock resources that are literally sitting on the floor, and, counterintuitively, incentivize our stewardship of it.

A Note On Industry and the Environment

Industrialization has dirty connotations. As you read the word, you might be picturing smokestacks, or worse, the soot-darkened faces of pre-pubescent factory workers.

I’d argue that the industrialization of new frontiers has been one of the most powerful engines of human progress. So I want to tell you how I think about industrializing the ocean while protecting it, because a desire to protect it, to restore it, is where I began this journey.

When we started Ulysses, we set out to help restore the oceans by using underwater robots to replant seagrass more efficiently. Seagrass meadows cover less than 0.2% of the ocean floor but store up to 18% of the ocean’s carbon, support roughly a fifth of the world’s fisheries, and have been declining at about 7% a year since the 1990s. This means we’ve lost roughly a third of all seagrass globally in the past few decades. Restoring these meadows is an important part of bringing back the vitality of the oceans.

That work is underway in Florida, Virginia, Western Australia, and in the Great Barrier Reef. As we waded into the ocean, however, we realized a few things.

First, we realized that the infrastructure we’d need to do restoration well didn’t exist. We’d have to vertically integrate and build most of it ourselves. Soon, we realized that the infrastructure we had to build was exactly what everyone else working in the ocean needed too. We’ll talk about that below.

Second, there are big differences between 19th-century industrialization and modern oceanic industrialization, the biggest of which is the difference between combustion and electric machines. Unlike those factories, or even modern oceanfaring vessels, our vehicles won’t emit smoke or smog. They run on electrons and don’t emit exhaust or greenhouse gases during operation. We will discuss this, too.

Third, we realized that when we expanded our economic ambitions, we also increased our environmental ones. I’ve just shown you what a pre-industrial ocean looks like. There is slavery, overfishing, the extinction of entire species, and ignorance of endless pots of gold. A pre-industrial ocean doesn’t mean an untouched ocean; it just means that the pirates are the ones doing the touching. Commerce brings law and order, if only selfishly, but that law and order has positive externalities. You are far less likely to be murdered in a Western Saloon today than you would have been in 1826. We believe that by industrializing the ocean, we are incentivizing its protection and restoration.

The ocean is too precious to leave to the pirates.

How to Industrialize a Frontier

So if we want to industrialize the ocean, how do we do it? There seem to be four stages that frontiers go through on the path to industrialization.

Discovery → Expedition → Access Breakthrough → Industrialization

Discovery proves the frontier is real and reachable. Expedition explores it, maps it, and finds the resources worth pursuing, but it does so episodically and expensively (or illegally). The fruits of expedition incentivize the development of the access breakthrough: a technology that dramatically collapses the cost of reaching and operating at the frontier, making entirely new categories of economic activity viable for the first time. That triggers industrialization: the frontier integrates into the broader economy, and industries emerge that couldn’t have existed at the old cost structure.

The American West went through all four stages. The early explorers proved it was there. For sixty years, wagon trains, fur traders, and the Oregon Trail explored it episodically. Then, in 1869, the Transcontinental Railroad met at Promontory Summit, Utah, collapsing the cost of access, and what followed was the industrialization of half a continent. The Homestead Act alone distributed 270 million acres, an area larger than France, Germany, Italy, and Spain combined. Western mines produced the copper that wired America’s cities, the gold and silver that backed its currency, and the timber and beef that fed its industrial workforce. Over the century that followed, the frontier compounded into trillions of dollars of real economic value and provided the physical substrate of the American century.

Space has followed the same arc. Sputnik proved it was reachable. Apollo and the Shuttle explored it at $1.5 billion per launch. Then SpaceX spotted that rocket materials cost about 2% of the selling price, vertically integrated them, built reusable boosters, and cut costs 20x. The space economy has tripled over the past two decades, reaching $613 billion in 2024, and is projected to hit $1.8 trillion by 2035.

Both of these examples are gross oversimplifications, but what I want to establish is that there is economic activity on the frontier even before industrialization. Risk-seekers are captivated by the promise of the untamed. But true market creation occurs when early economic activity incentivizes and funds access infrastructure, and access infrastructure facilitates normal business creation. Industrialization moves frontiers from limited, risky, low-volume, potentially high ROI trades to higher volume, safer, lower-individual-ROI but higher-overall-output activities.

Today, the ocean’s pre-industrial economy is already enormous. We spend $2.6 trillion a year across shipping, offshore oil, fishing, subsea cables, and coastal ports. That is an order of magnitude larger than the space economy was when SpaceX was founded.

But the economics of the ocean severely limit where we can operate, and therefore the size of the ocean economy.

We ship across the surface because the surface is the cheapest layer to traverse, and the only one where our GPS and WiFi work. We drill oil from expensive fixed platforms because oil is valuable enough to pay for the helicopters and the two-week crew rotations. We fish from the top of the water column because that is where our tools reach, and as a result, we decimate the stock.

In short, we are limited to the easiest-to-access parts of the ocean, unless an extremely valuable and fairly predictable commodity can justify deeper exploration, just as the fur trade did during America’s westward expansion.

To the extent that we can decrease the cost of access, we can increase the size of the ocean economy. We need the ocean’s railroad. It is surprising that, if $150 billion of activity was enough to trigger a railroad for space, $2.6 trillion of pre-industrial economic activity didn’t trigger one for the ocean long ago.

Why have we not built the ocean’s railroad?

The Ocean Fights Back

Two factors must be simultaneously present for industrialization to occur: suitable technology and urgent demand.

Humans have had boats for a very long time. As early as 3000 BC, the Austronesian people migrated throughout the islands of the Indo-Pacific using sailboats like this one.

But the ocean has never made traversing its surface easy. During the Age of Sail, an estimated 3-5% of merchant fleet ships were lost per year. Rather than being retired, most wooden ships ended their careers by being wrecked, foundering, or being lost at sea. Over time, we got better at making sturdier boats. But even still, the 1990s’ most popular movie was about the sinking of the “Great Unsinkable” Titanic.

Stewart Brand recently published Maintenance with Stripe Press. He chose to open the book with the story of the 1968 Golden Globe Race, the first solo, non-stop circumnavigation of the earth by sailboat. The race’s sponsor, The Sunday Times, charged no entry fee and laid down almost no rules. Sailors simply had to leave from a British port, travel around the globe, and come back to it without stopping.

Nine sailors entered the race, and historians mainly focus on three of them.

Donald Crowhurst, a brilliant inventor with arguably the most technologically advanced boat in the race, never made it past the Atlantic. His electronics failed, his hull leaked, and the isolation broke him. He falsified his logbook, drifted in circles, and eventually stepped off the back of his trimaran into the sea. His body was never found.

Bernard Moitessier, the most gifted sailor in the fleet, abandoned the race despite being far in the lead. Months alone in the Southern Ocean convinced him he’d rather keep sailing to Tahiti than return to civilization.

Robin Knox-Johnston spent 312 days at sea, and most of them were spent not sailing but repairing. Where the others designed their sailboats for technological superiority or speed, Knox-Johnston designed his to be fixed:

To prepare SUHAILI for a ten-month passage, most of it in the world’s roughest waters, he packed into his small boat all the ‘materials and tools’ he could imagine he might need – specialized wrenches for every exotic nut on the boat; ditto for screwdrivers; a sailmaker’s bag full of needles, sewing palms, and twine; a bosun’s bag with every kind of shackle, thimble, and marlinspike for managing all his steel wire rope; a spare bilge pump and extra rubber pipe; 12 yards of canvas; caulking chisels and cotton; plenty of oil, glue, and Stockholm tar; spare parts for everything mechanical; and medical supplies for repairing himself.

It was the right move, because he spent most of his 312 days on the water fixing. And yet… “I realized I was thoroughly enjoying myself,” he said later. Still, the enjoyment was the enjoyment of a man who understands that the ocean is in a permanent state of war with anything humans put in it.

Knox-Johnston won the race, “and the prize of £5,000 – which he gifted to Donald Crowhurst’s bereaved wife and young children.”

Typically, the story is told as a story of different flavors of the human spirit, and it is certainly about that. But Brand chooses to tell it as a story about maintenance. The universe tends towards entropy, but it does so particularly quickly on the open sea. There is no environment on earth that so persistently attacks those who dare challenge it.

Which is to say that the ocean is the hardest frontier on Earth. It fights you every day that you’re in it. The ocean doesn’t necessarily want to fight you. It is simply a matter of the ocean’s constitution.

Saltwater is one of the most corrosive environments on earth, eating through steel, degrading composites, and attacking every electronic component it reaches. Since water flows happily through cracks, it ultimately reaches almost everything. The North Sea, one of the most developed ocean regions on Earth, is so punishing that corrosion alone accounts for roughly 60% of maintenance costs on production platforms, and lifetime operating spend routinely exceeds the original construction cost. Offshore wind has struggled to meet expectations because maintenance costs 2-3x more than onshore wind, components break down more frequently, and turbine performance degrades by an average of 4.5% per year. The global cost of marine corrosion runs to $50 to $80 billion a year, and that’s just for the structures we’ve bothered to build.

Then there’s biofouling (think barnacles): the moment you put something in the water, organisms begin to colonize it. Within weeks, without active maintenance, a clean sensor is blind, and a clean hull is dragging tons of extra weight. The offshore oil industry spends billions per year fighting exactly this. Everyone who chooses to operate on the ocean must become a Robin Knox-Johnston.

Below the surface, the battle only intensifies. Pressure increases by one atmosphere every ten meters of depth. At the average seafloor, you’re looking at 370 atmospheres. Everything on a subsea vessel must be engineered for forces that have no analog on land or in space, and which vary continuously with depth. Space is extreme, but the vacuum is a single, well-understood engineering problem. In the ocean, every meter deeper you go changes the engineering envelope.

We have thus far described fair weather conditions, but the ocean is a stormy place. Storms routinely destroy purpose-built infrastructure: Hurricanes Katrina and Rita alone destroyed 113 offshore platforms and ruptured 457 pipelines.

In each one of these cases, unless you are right there with your vessel, you’re unlikely to even know when things go wrong, because the ocean is a communications desert. GPS doesn’t work underwater at all. Radio and optical signals penetrate about 20 meters before the water absorbs them. Acoustic signals can travel further, kilometers in some cases, but at painfully low bandwidth, about a million times slower than surface 5G. To get around these constraints, you can tether a vehicle to a surface ship with a cable, but that limits your range and requires an expensive crewed vessel overhead, at which point, you’re still stuck in the expedition model.

The ocean is a domain that degrades everything you put in it, crushes anything you send deep, destroys what you bolt to the surface, and isolates whatever survives.

This is before we even address the immense scale of the ocean. So even if you solve every engineering problem, you still face the question of how to cover a domain that dwarfs anything we’ve ever tried to operate in while fighting corrosion, biofouling, pressure, storms, and the comms desert. None of these is impossible to overcome, but every one of them is a tax. A tax currently paid in steel thickness, redundant systems, $20,000/day ship time, mobilisation windows, insurance premiums, and in the engineers you have to send offshore to fix what can’t be fixed from shore.

And on top of that, the industry that grew up around this domain never got what aerospace and automotive got: scale, software, modern supply chains. Maritime is still largely a cottage business of bespoke parts and hand-built systems. So you pay the tax twice: once to the physics, and again to the pre-modern industry that evolved to serve the physics.

We call it the Ocean Tax: the compounding cost the sea, and the industry that grew around it, extracts from anyone who tries to do anything in it. Every existing ocean company is, at heart, a machine for paying the Ocean Tax.

So the natural question is, how do we get around it?

What It Would Take to Industrialize the Ocean

The good news about the ocean being such a challenging environment is that every problem that the ocean throws at you doubles as a design requirement. If we solve for the challenges, we know what kind of system to build.

The ocean is vast. Instead of a few expensive, exquisite vessels, you need a lot of cheap ones.

This is the same trend that we’ve seen in space, going from a few billion-dollar communications satellites to thousands of small, cheap ones. The latter means that the network can cover more area more reliably than single satellites can, and inevitable damage to a single satellite doesn’t knock it out. It is resilient. We are seeing the same trend in defense, with the move towards high-volume, attritable drones instead of a few exquisite platforms.

Like space, it is hard to comprehend just how much room there is to cover in the ocean. The 361 million square surface kilometer number hides just how big it is, because it also goes deep. Industrialization requires coverage of the X, Y, and Z axes. The ocean is more than two miles deep, on average, and its deepest trenches are so deep that Mount Everest could be dropped into the Challenger Deep with 2km of clearance above it. Its total volume is something like 1.3 billion cubic kilometers, which is so large that I’m not even sure what to tell you, other than to say that the number of vehicles we put in it each year stands absolutely no chance.

Today, the entire global autonomous underwater vehicle (AUV) market produces roughly 1,000 vehicles per year, and the incumbents’ AUVs cost on average $500k to $5 million each. For scale, a thousand units across the entire ocean is equivalent to just 27 vehicles across the entire United States. Imagine trying to patrol the entirety of the United States with just 27 vehicles. And that is just the surface comparison! These vehicles are precious, and are treated as such, which is no way to build an economy.

We think the answer to making ocean infrastructure possible is manufacturing at radically lower costs. At Ulysses, we build AUVs for as little as $50,000 per unit, a 10x to 100x reduction over the most commonly sold incumbent models. At that price, we and our customers can think in terms of fleets rather than individual vehicles, and fleets, we believe, are the unit of infrastructure the ocean demands.

The ocean demands you operate at the surface and subsea as a single integrated system.

How do you design a vessel meant to operate on the surface and in the deep ocean, when the two are almost entirely separate domains, with radically different physics? Well, you don’t.

You design a system that handles both, consisting of separate vehicles optimized for each. The surface is where GPS and Starlink work, and doesn’t face the same pressure challenges as lower down, so that’s where communications, fuel, power, and logistics live. But most of the work is below the surface, where the undersea cables, seafloor, and marine ecosystems lie. You need vehicles that can go where the work is, and you need surface platforms that can deploy, recover, recharge, and connect them to the rest of the world.

A fully autonomous system is a much cheaper system, because the cheapest AUV in the world is still expensive if it’s hand launched and recovered, and requires a crewed ship to operate it. There’s a joke in the maritime industry that “unmanned” systems are only unmanned in the sense that the people aren’t on or in them, but right beside them while they operate. As long as humans have to babysit autonomous vehicles, they’re not really autonomous, and it will be impossible to achieve scale.

We’ve designed the system at Ulysses with those requirements in mind. Our underwater vehicle Mako goes deep and does the work. Our autonomous surface craft and mothership, Leviathan, equipped with our autonomous launch, recovery, and recharge platform, Kraken, remains on the surface and serves as the fleet’s connective tissue: deploying Makos, recovering them, recharging their batteries, and relaying their data to onshore operators via satellite. Leviathan is a working asset in its own right, a node in a domain awareness network that can carry its own sensors and serve its own missions. Together, linked by Kraken, they form a single integrated system that can persist at sea without retreating to port.

The underwater environment demands autonomy.

Because of the underwater communications desert we described earlier, you cannot supervise an underwater vehicle in real time the way you would a drone in the air. The bandwidth simply isn’t there, and the latency would make remote control dangerous and unreliable.

AUVs need to think for themselves. Because we’re talking about large fleets of small, inexpensive vehicles, there is no room in either the cost structure or the vehicle itself for a human pilot. Each AUV needs to navigate without GPS, make decisions based on what they see, avoid obstacles, adapt to changing conditions, and execute complex missions with only periodic check-ins.

This is as much a compute problem as it is a software one. At Ulysses, we pack over 100 times more onboard compute than anyone else in our underwater form factor, data-center-class GPUs running inside an AUV. Since we can’t talk to our vehicles, they need to be smart enough not to need us.

Industrializing the ocean demands both observation and action.

Most ocean technology stops at sensors, which are valuable in their own right. However, sensors are not by themselves industrial infrastructure. To industrialize a domain, you have to be able to act on what you observe.

Consider a vehicle that can only observe. It would swim around, notice that a pipeline was in need of repairs, ascend to the surface, and send a message back to a team of humans, who would then schedule and weather permitting, get on an expensive vessel, travel out to the site of the damage, and send a human or ROV (remote operated vehicle) below to make the necessary repairs. That process costs an unnecessary amount of time and money. Conversely, a platform that can both observe and act would be able to both inspect and repair the pipeline, quickly and cheaply. More persistent monitoring would also mean that each individual repair is likely to be simpler.

Here, modern robotics is key. The fleet of machines we send into the ocean must be able to manipulate, intervene, and repair, which means that they need arms, tools, and the dexterity to do meaningful work at depth. If they can, you’ve turned a sensor network into an industrial presence. This is the hardest part of the stack to get right, but given our heritage in sea grass planting, it’s the part that we started with.

The ocean demands a fully integrated system.

You may have noticed, reading this section, that each requirement builds on the others. If you want low costs, for example, vehicles need to be autonomous and they need to be able to act. As soon as you introduce humans back into the equation, coverage and persistence drop while costs rise. If you want vehicles that are truly autonomous, persistent, and can act at depth, you need surface vehicles and subsea vehicles that work together to do what each can do best.

We think this is how you industrialize the ocean, and we believe that the solution has to look more like a network of Starlinks than a railroad: a distributed network of coordinated assets, surface and subsea, each playing a different role. Together, they collectively give you persistent coverage of the hardest domain on earth and the ability to do something about what you find.

This is not a novel insight. Mariners have understood it in theory for at least a few decades. But there have been two challenges: it wasn’t possible to build, because the underlying technologies weren’t ready, and it wasn’t easy to fund, because the urgent demand for such capabilities didn’t exist.

Within the past five years, both of these aspects have changed.

Why Now?

There is perhaps no more important question that a technology startup can answer than “why now?” I argued earlier that every frontier becomes industrialized when supply-side technology and demand-side urgency converge. For the ocean, that convergence is happening right now.

Supply Side: The Ocean Stack

In Cable Caballero, Packy coined “Curve Convergence“: if multiple component technologies improve on exponential curves as industry architectures remain frozen around outdated assumptions, it’s possible to create the opportunity for disruption with a better product.

The ocean is experiencing its Curve Convergence right now. Four exogenous technology curves all crossed critical thresholds in the same window, roughly 2020 to 2025. Each crossing has removed a specific barrier that previously made persistent ocean work impossible, and together they enable a total rewrite of how mankind can work at sea.

Connectivity

Before Starlink, communicating at sea meant paying over $5,000 a month for slow, unreliable satellite connections with multi-second latency. That meant any “autonomous” ocean vehicle was really just pre-programmed and running a script with no ability to adapt in real time.

Starlink Maritime changed this completely: it offers roughly $250 a month for 100+ Mbps with sub-70-millisecond latency, or a 100x improvement in price-performance in about three years.

Starlink gives you high-bandwidth, high-fidelity visibility into what your fleet is doing, at a cost that scales. Before, you might have gotten a trickle of compressed data over a satellite phone. Now, if one of your vehicles surfaces or relays data through a surface platform, you’ll get full sensor feeds, video, vehicle health, and mission statuses. The full picture will be streamed in real time to an operator who might be sitting in Dublin or San Francisco. The cost also helps: $250 a month instead of $5,000 or more has meant that connectivity can be placed on every surface asset in a fleet, making it possible to coordinate hundreds of vehicles across an ocean basin. Underwater, the vehicles can think for themselves, and when they’re back above ground, you can see everything they saw.

Energy

Energy is one of the most critical binding constraints of ocean work. Vehicles need to stay on or under the water for days, weeks, or indefinitely, and that requires generating, storing, and efficiently using energy in ways that simply weren’t possible a decade ago. Improvements have come on those three fronts simultaneously.

On generation, solar costs have fallen 75% in a decade, and that drop has reached the water. Solar- and wind-powered surface vehicles have already stayed at sea indefinitely on harvested energy alone. Below the surface and beyond solar, wave energy is reaching viability, achieving capacity factors around 90% (compared to 30–40% for offshore wind and 25% for solar) because waves run round the clock.

On storage, battery energy density has doubled in a decade, from 150 Wh/kg in 2015 to over 300 Wh/kg today, while pack costs have collapsed from $1,191 per kWh in 2010 to $115 per kWh in 2024. As Not Boring readers will understand intimately, increased battery density is important for everything from drones to AVs, but nowhere is it more important than it is underwater: every increment of energy density directly extends mission duration. Five years ago, a subsea vehicle on the best available batteries could run four-hour sprints before it needed to be recovered and recharged. Today, double the energy density, dramatically cheaper packs (which means you can afford to carry more total capacity), and more efficient motors and drive systems combine to make multi-day deployments possible.

All of these improvements compound. Each one matters, but it’s the combination, more energy per kilogram, cheaper per kilowatt-hour, less energy consumed per kilometer, that transforms the operating model from sprint-and-retrieve to deploy-and-sustain.

Compute and AI

The vast majority of underwater work today is still performed by remotely operated vehicles (ROVs), tethered to a surface vessel by an umbilical cable and piloted in real time by a human operator watching a screen. The ROV goes where the pilot tells it, sees what the pilot looks at, and does what the pilot commands. It’s capable and proven, but it means every subsea operation requires a crewed ship overhead, which is why day rates for subsea work start at six figures.

Truly Autonomous Underwater Vehicles exist, but their capabilities have been narrow. They can run pre-programmed survey routes and collect data (sonar mapping, pipeline inspection, environmental monitoring), but that’s essentially read-only work. When it comes to acting on what they find, making a decision, adjusting course, manipulating something physical, the vehicle surfaces and a human takes over. The ocean has been a domain where machines can look but not think, and certainly not act.

That’s changing fast as the AI-driven terrestrial compute bonanza reaches the water. Onboard processing power has increased by orders of magnitude in the past five years, and the cost of edge inference is plummeting. Vehicles can now run real-time object detection, navigate around unexpected obstacles, and make mission-critical decisions without surfacing for instructions. The trend is moving from read-only to read-write: not just gathering data but interpreting it and acting on it autonomously. A thousand vehicles can’t each have their own human pilot. They need to think for themselves, and for the first time, they can.

Motors and Sensors

An underwater vehicle is, at its core, a collection of motors, thrusters, actuators, sensors, seals, and structural components held together inside a pressure housing. Until recently, every one of those parts was bespoke, sourced from specialist defense suppliers who charged accordingly. This is one of the reasons that incumbent AUVs built by legacy defense contractors with artisanal production methods cost so much.

What has changed recently is that adjacent industries built new supply chains for us. The commercial drone industry commoditized motors, ESCs, and flight controllers. The EV industry drove down the cost of power electronics, drive systems, and battery management. Consumer electronics made high-performance sensors and embedded systems cheap and abundant. Packy has written about this phenomenon as the modern electric tech stack: the convergence of commodity components that lets you build sophisticated electromechanical systems at a fraction of what they used to cost.

A new generation of ocean vehicles, built on these supply chains, can be produced for hundreds of times less than incumbent vehicles.

That cost collapse makes the fleet model possible. We couldn’t possibly cover 361 million square kilometers of ocean with million-dollar vehicles, but we can with fifty-thousand-dollar ones. Thanks to these curves, we can make cheap vehicles useful enough that we need a lot of them, and by building more of them, we will make them even cheaper and more useful.

All four of these curves had to converge for ocean industrialization to become possible, and finally, they have. It’s the convergence of what I call the Ocean Stack that makes the phase transition possible.1

These curves are a thing of beauty, and we are fortunate to be alive at a time when so many of them are converging at once. I view it as our role to combine the Ocean Stack into useful products that are better, faster, cheaper, and entirely different in capability than previous generations of hardware.

Demand Side: Everything, Everywhere, All at Once

No matter how good the technology, in order to reach the scale that both affordability and industrialization require, the supply needs to be met by demand, which it currently is, from all angles. What makes right now different from every previous decade is that the demand drivers are independent of each other. Any one of them would justify building ocean infrastructure, but all of them are accelerating simultaneously.

Undersea Cables

Ninety-nine percent of intercontinental internet traffic travels through submarine fiber-optic cables. Over a million kilometers of cable sit vulnerable on the ocean floor with essentially zero persistent surveillance. Now, there is simultaneously more demand for new cables and greater threats to them, natural and deliberate.

First, AI is driving an explosion of new builds: Meta announced Project Waterworth in 2025, a 50,000-kilometer cable spanning five continents --- the longest in history.

Amazon, Microsoft, and Google are all commissioning major new transoceanic routes. Investment in subsea cables is expected to nearly double between 2022 and 2027, with tech giants now controlling roughly half the market.

Second, those cables break 100 to 200 times per year due to accidental damage --- trawlers, anchors, earthquakes, and sharks --- and each break requires expensive, slow, ship-based repairs with no persistent monitoring in between.

Third, and most urgently: deliberate sabotage is escalating. In November 2024, two Baltic Sea cables were severed simultaneously --- the Lithuania-Sweden interconnect and the 1,200-kilometer C-Lion1 linking Finland to Germany. A month later, on Christmas Day, a Russian shadow-fleet tanker dragged its anchor through the Estlink 2 power cable and four telecom cables, cutting Estonia’s cross-border electricity capacity by two-thirds. NATO launched Operation Baltic Sentry in January 2025, sending frigates, patrol aircraft, and naval drones because no other assets were available to detect or deter these attacks.

Just two weeks ago, the UK accused Russia of running covert submarine operations over its cables and pipelines. Addressing Vladimir Putin, UK Defence Secretary John Healey said, “We see you. We see your activity over our cables and our pipelines, and you should know that any attempt to damage them will not be tolerated and will have serious consequences.”

In short, more cables are being built, more value is flowing through them, and more bad actors are learning what sharks have known since humans started laying subsea cables: that, Healey’s warning aside, they’re basically unwatched and undefended.

Today, subsea cable monitoring and repair is a ~$3B market expected to double by the early 2030s. That, however, underestimates the size of the opportunity, because the market is constrained by vessel supply, not demand. There are something like 60 specialized repair vessels in operation worldwide, and proactive monitoring is still in its infancy, as are novel approaches to using the cables themselves as sensors in a method called Distributed Acoustic Sensing (or DAS). Lower the cost of supply, and we will expand the market.

Critical Minerals

The Clarion-Clipperton Zone in the Pacific alone contains more cobalt than all known land reserves, along with polymetallic nodules rich in nickel, manganese, and copper.

Providentially, these are the exact minerals that EV batteries, wind turbines, and AI hardware are consuming at accelerating rates. Regulation, along with cost and technology, has kept these from being mined, but the regulatory dam seems to be cracking. After years of stalled negotiations at the International Seabed Authority, The Metals Company forced the issue in 2025 by filing the first-ever application for a commercial deep-sea mining license, bypassing the ISA entirely via a US regulatory pathway established in 1980. Whether you think deep-sea mining is the answer or not, the pressure to access these resources is intensifying, and you cannot extract, survey, or even responsibly monitor extraction without ocean infrastructure that does not yet exist.

It is not worth trying to pin down real numbers here, because the field is so nascent and the numbers so large. The Metals Company estimates the NPV of its first two projects to be $24 billion. The U.S. Geological Survey study on polymetallic nodules projects that if deep-ocean mining follows the evolution of offshore petroleum production, about 35–45% of the demand for critical metals will come from deep-ocean mines by 2065. And there is an estimated $233 trillion worth of value in polymetallic nodules worldwide, although, as with asteroid mining, these reserves will be difficult to access and, if accessed at scale, should dramatically decrease prices. For our purposes, it is safe to say that there is a lot of value down there.

Defense

When we started discussing writing this piece, maritime defense was already an important topic. We’ve all seen the chart comparing China’s shipbuilding capacity to the United States’. The ability to manufacture ships, traditional or autonomous, big or small, will be a deciding factor in any conflict between China and the U.S. Ideally, this ability will also be instrumental in deterring conflict in the first place.

Over the past couple of weeks, however, the question of maritime defense has become urgent. I’m writing this in April 2026, as the IRGC and United States are locked in a conflict over whether to reopen the recently closed Strait of Hormuz, the passage through which a fifth of the world’s oil supply normally flows. When Iran first closed the Strait, tanker traffic dropped 70%. On the surface, ships passing through the Strait have been struck by missiles, and below the surface, Iran has planted mines. In a dark confirmation of everything we have been arguing in this piece, Iran was unable to comply with President Trump’s demands to allow more ships to pass through the Strait, because it is unable to find the mines it just recently put there.

In the Red Sea, Houthi attacks forced global shipping to reroute around Africa, adding 11,000 nautical miles and 10 to 15 days per voyage, at a cost of roughly $1 million per ship. Container rates on the Shanghai-to-Europe route quadrupled. A trillion dollars worth of goods were disrupted in the first six months of the Houthi effort alone.

Meanwhile, China is building submarine-sized autonomous underwater drones with ranges exceeding 18,000 kilometers, developing cable-cutting tools that operate at a depth of 4,000 meters, and constructing an “Underwater Great Wall” of seabed sensors across the Indo-Pacific.

The post-war international order was fundamentally underwritten by the United States’ ability to make the ocean safe for trade. That either has broken or is breaking, depending on who you ask.

The United States is aware of it and is putting resources behind maritime superiority in the modern paradigm. The FY2026 Pentagon budget allocates $13.4 billion for autonomous systems. This is the first time autonomy has been its own budget line. The Navy alone gets $5.3 billion, a 70% year-over-year increase.

This is another truth about frontier industrialization. Every frontier in history was industrialized when powerful competition demanded it.

America pushed West to fulfill our Manifest Destiny. We raced Russia to the Moon to achieve superiority in the Cold War. The drive to industrialize the ocean is now fueled by China, Iran, and a world waking up to the fact that America is no longer capable of guaranteeing the maritime safety that has underpinned global trade since World War II.

Beyond the specific demand drivers, though, there is something more fundamental driving the Oceanic Renaissance.

If you look at the sweep of human history, there is an inexorable, consistent drive to pursue the frontier. We make the most of everything. We find everything. Technology, capitalism, human ingenuity, creativity, and sheer stubbornness combine to produce extraordinary outcomes from the resources available to us. Every frontier that could be industrialized eventually was.

The ocean is the last frontier. The technology is finally ready at the same time as demand is coming from every direction at once. This is going to happen. The only question is who builds it and how fast.

The New Ocean

One thing is certain: the incumbents cannot lead this transition.

Other than the water, the ocean industry looks like any other that’s dominated by sclerotic incumbents too content with the world that was. You’ll even recognize many of the names.

Maritime defense primes include old faithfuls like Lockheed Martin, RTX, L3Harris, Leidos, and Thales. They sell everything from combat systems to sonar suites to legacy AUVs, often in a cost-plus model. Shipbuilders, like HII, General Dynamics Electric Boat, BAE Systems Maritime, Kongsberg, and the big three Korean yards, turn out hulls on multi-year contracts, which are cost-plus as well. Then there are companies that you’ve probably never heard of, the subsea service primes like Fugro, DOF, Oceaneering, Subsea 7, and ASN, which run thousands of campaign-based vessels for oil majors and telcos.

Between the ocean’s incumbents, hundreds of thousands of people generate tens of billions of dollars a year, every cent of it organized around the assumption that the ocean is expensive and will stay that way. You can forgive them for operating as if nothing will change, because that’s how their customers buy.

This is textbook Innovator’s Dilemma.

Christensen’s point is that incumbents fail by doing everything right for their best customers. Here, ocean incumbents’ best customers (navies, oil majors, telcos, shipping companies) are conservative, risk-averse, and demanding. They want proven, crewed, campaign-based services with mil-spec redundancy and rigorous SLAs. Autonomy-first, priced-for-scale platforms don’t meet those requirements yet. So the incumbents, rationally, keep investing in the fleets they already own in order to protect their margins and serve their best customers better.

By the time the new cohort is good enough to matter, the incumbents will have a decade of sunk cost in crewed vessels, a specialist workforce that has never shipped an autonomy stack, a mil-spec supplier network that cannot pivot, and an entire organization whose every incentive points at the existing business. The incumbents never lose because they’re stupid, or even because they don’t listen to the market. In fact, they listen to the current market too closely.

Once again, space is the case study. For fifty years, launch was owned by a small cohort of cost-plus primes – Lockheed, Boeing, and Northrop Grumman – whose economics depended on expendable rockets, government contracts, and scarcity. Then a new generation showed up, looked at the same physics, and decided that if you vertically integrated the supply chain, built for reusability, and priced for scale instead of scarcity, everything changed. That cohort, led by SpaceX and including Rocket Lab, Astranis, Planet Labs, Relativity, Varda, K2, and far too many to list, has a name now: New Space. As the cost of a kilogram to orbit fell by more than an order of magnitude, entire industries, including Earth observation, satellite-IoT, constellation broadband, and even space drugs, space-laser energy, and orbital data centers became possible and potentially profitable.

The same thing is happening in the ocean right now.

A new generation of companies is rebuilding the ocean from first principles. They operate less like modern versions of the rusty old ocean incumbents, and more like New Space companies that happen to work in salt water instead of vacuum.

While each of these companies is different, they tend to share some common characteristics:

• Vertically integrated instead of subcontracted.

• Autonomy-native instead of crewed-first with autonomy bolted on.

• Software-defined instead of hardware-frozen.

• Built on commercial supply chains (drone motors, EV battery packs, Jetson-class compute) instead of custom mil-spec.

• Priced for scale instead of for scarcity.

We call this cohort the New Ocean.

As with any market map, there are a number of ways that we could slice and dice our fellow ocean startups, but the simplest is to group them into three buckets: infrastructure, platforms and vehicles, and vertical applications.

At the foundation are the Infrastructure companies that enable the Ocean Stack. These are the companies that are bending the curves on which the rest of us ride.

Starlink put a high-bandwidth, low-latency pipe to every square meter of ocean for the price of a weekly Uber habit. Panthalassa will generate persistent electricity at sea from wave energy, a marine renewable that runs through the night at much higher capacity factors than solar or wind.

Karpowership already operates forty floating power stations delivering 7.5 gigawatts across multiple continents, proving that energy supply at sea can be mobile and matched to demand.

Armada is deploying full compute clusters at sea, bringing the data center to the data with nearly free cooling from the ocean itself. Kraken Robotics builds the synthetic-aperture sonar that turns a cheap AUV into a survey-grade mapping platform. Each new capability that one of the Infrastructure players provides is a capability that the rest of the Ocean Stack can snap in and use.

In the middle of the stack are the Platforms, or Vehicles. These companies, including Ulysses, take advantage of the infrastructure layer and enable vertical applications. Collectively, these are the railroads of the ocean’s industrialization.

Above water, Saildrone has sailed more than two million autonomous nautical miles across the world’s oceans on solar and wind power alone. They are proof that persistent surface presence without humans or fossil fuel is an operational reality.

Saronic is rebuilding American shipbuilding around autonomous surface vessels for defense, attacking the same capacity gap that keeps the US Navy dependent on decades-old hulls. Blue Water Autonomy is designing 190-foot Liberty Class uncrewed naval ships from a blank sheet. The ships would be larger than any USV afloat today, built for the open ocean rather than the coastline.

Regent is electrifying coastal transport with all-electric seagliders that skim above the water faster than a ferry and don’t need an airport. Poseidon Aerospace is flying autonomous cargo aircraft to move freight across ocean basins at a fraction of the cost of a container ship and a fraction of the time of a plane.

Navier is building electric hydrofoiling boats that use 75% less energy than traditional hulls. Arc is rethinking how boats are built entirely: electric-first, hull-up, designed for modern manufacturing instead of inherited from the combustion era. In addition to its electric sport boat (which is incredibly fun, fast, and quiet), Arc is now moving into commercial tugboats on the same electric platform.

Beneath the surface, Anduril’s Dive division is producing the Dive-LD large-displacement AUV and leading Australia’s Ghost Shark program, bringing a defense-tech mindset to underwater warfare.

Across the sensor layer, Andrenam is weaving low-cost AI-powered sonar buoys into persistent maritime domain awareness meshes to string together a distributed listening grid from surface to seabed. Sofar Ocean is doing something similar for ocean observation, operating the largest privately-owned network of ocean sensors on Earth and turning real-time wave, weather, and current data into a platform the rest of the stack can build on.

With modern Infrastructure and Platforms, new Vertical Applications become economical.

The Metals Company and Impossible Metals are making the case that the ocean floor holds the cleanest path to the critical minerals the energy transition demands. TMC is filing the first-ever commercial deep-sea mining license, while Impossible builds robots that selectively pick nodules off the seabed rather than vacuuming it.

Shinkei Systems is using robotics and computer vision to automate the ancient Japanese ike jime harvesting method. They put their machines right on fishing boats to instantly, humanely, and precisely kill fish, eliminating the stress response that degrades most of the world’s wild-caught seafood and tripling its shelf life.

A market map of the New Ocean today looks platform-heavy, because that’s what new frontiers looks like at this stage. The platforms always show up early. But even among the platforms today, we think something is still missing. Each company on this map is solving a critical slice, but to truly industrialize the ocean — to turn it from a domain of campaigns and expeditions into one of systematic, repeatable, economically productive work at scale — you need all of those pieces operating as one integrated system: persistent surface and subsea vehicles, edge autonomy, real-time communications, and the ability to act. You need the whole stack in one.

That’s our thesis at Ulysses, and that is what we are building.

We’ve always believed that the ocean was Earth’s most mismanaged resource, and if it had better tools, it would be transformative for society. We looked at how SpaceX had approached space — the bet on vertical integration, the bet on pricing for scale instead of scarcity, the refusal to accept that the physics of the domain justified the legacy cost base — and decided the ocean needed someone willing to take the same approach. Our bet is that if the whole stack is built together, as one autonomous system, you can dramatically collapse the cost of working in the ocean.

My co-founders and I think about the opportunity through a single governing metric: the cost of ocean work, which we define as the cost of active operation over one square kilometer of ocean for one month ($/km²-month).

We believe the ocean is a constrained economy, severely limited by the tools available to work in it. Every constrained economy has a governing metric, the one number that, if you move it, everything else follows. For space, it was $/kg to orbit. For the ocean, it is the cost of ocean work.

That cost is unreasonably enormous today: covering a meaningful area of ocean with incumbent methods (crewed vessels, campaign-based operations, six-figure day rates) costs millions of dollars per month. If we can collapse it, we’ll make it economically viable to monitor cables, survey mineral deposits, manage fisheries, patrol shipping lanes, and do a hundred things nobody has thought of yet, at a scale that was previously impossible.

The ocean economy today is $2.6 trillion, built almost entirely on transit, extraction, and coastal adjacency. Imagine what it becomes when you can actually afford to work in the ocean, persistently, at scale. We are very early on the cost curve.

We collapse the cost of ocean work in two ways: on OpEx and on CapEx.

On OpEx, we integrate surface and subsea vehicles into a single autonomous system, removing the crewed surface vessel that drives the majority of incumbent operating costs. Our surface platform, Leviathan, deploys, recovers, and recharges our subsea vehicles via our autonomous launch, recovery, and recharge platform, Kraken, without returning to port and without requiring a crew of 80 to 150 people on board.

On CapEx, we design and manufacture the vehicles ourselves. Our underwater vehicle, Mako, is a modular AUV that starts at $50,000 per unit, compared to $500k to $5 million for incumbent AUVs. It carries more onboard compute than any small or medium AUV on the market, and we build for read-write, not read-only: these vehicles are designed for robotics from the start, not just data capture. The modular design means we and our customers can reconfigure the same vehicle for entirely different missions in minutes.

We’re deeply vertically integrated because we have to be. The existing supply chains and component manufacturers for ocean hardware are slow, expensive, and not built for scale. So we design and build many of our sensors in-house, we do our own manufacturing, we do our own PCB assembly, we develop all our own software from the autonomy stack down to the middleware, and we’re progressively bringing more complex components (navigation systems, advanced sensors, motors) under our own roof. And we believe the best way to capture the value of collapsing the cost of ocean work is to sell the work itself: we operate these vehicles for customers as a service, selling outcomes rather than hardware.

Build the fleet, deploy the fleet, sell the coverage.

The demand thus far has validated our thesis. Just this month, a major commercial customer requested a network of 10,000 vehicles operated as a service. Another requested at least 1,000. We have never seen numbers like this before, and they keep getting bigger. I fully expect this year to be the year demand outstrips our ability to supply, and our challenges shift from market creation to manufacturing and scaling.

Scale, after all, is what it will take to truly industrialize the ocean.

What an Industrialized Ocean Looks Like

We’ve drawn many parallels between the ocean, the American West, and space in this essay. We expect that there will continue to be many similarities between those stories and the one that plays out at sea. Permanent infrastructure will replace episodic missions, and it will enable new types of businesses to operate at high volume and affordable prices. Governance will replace lawlessness, because those businesses will have assets they need to protect.

Space and the ocean share the same incumbent problem. Their incumbents are incentivized to keep things the way they are, which shows up as a high cost-physics gap, or what Elon Musk calls the “Idiot Index.” Prices are too high for all but the most profitable or profit-agnostic use cases. The same strategy to beat the incumbents applies to both, as well: vertically integrate and collapse the supply chain, and sell what used to be expensive hardware as an affordable service.

But the ocean is unique. Where space inspires sci-fi, the ocean inspires poetry. And the ocean has structural advantages that space doesn’t.

The ocean economy is already seventeen times larger than the space economy was at its inflection point. The ocean contains physical resources (minerals, energy, protein) that we have not yet been able to access (or find) in space. And the regulatory environment is the most permissive of any frontier: international waters begin 12 nautical miles from shore, with no FAA equivalent beyond that. The ocean is where modern cowboys roam.

The exciting part about operating among the cowboys is that it’s hard to predict exactly how things are going to play out. But I have a few thoughts and predictions.

Permanent autonomous infrastructure will replace episodic campaigns. Fleets of autonomous vehicles will operate continuously across ocean regions, persisting there as Starlink satellites do in orbit. Docking and charging stations will create underwater and surface logistics networks. Energy nodes will harvest wave and solar power. Compute clusters will float at sea. The ocean’s infrastructure will be networked instead of physically connected, but it will be no less permanent.

Resource management will move from blind to precise. Fishing will evolve from blind harvesting to precision management, replete with real-time stock assessment, targeted harvesting, and sustainable yields at scale. Mineral extraction on the seafloor will become viable with proper monitoring. We will harvest from the ocean itself at an industrial scale while stewarding its resources. The resources have been there since long before we were, but it’s only now that humans are cracking the economics of managing and accessing them with technology, as we once did with agriculture to great success in the Green Revolution.

Governance will follow infrastructure and economy. Sunlight is the best disinfectant, they say, and they haven’t even seen what sunlight does to a surface covered with salt water. The era of lawlessness will end through better visibility. The Bad Guys can’t fish illegally if the good guys can track every vessel, can’t cut cables if the routes are patrolled, and can’t dump waste if the water chemistry is monitored. The Wild West tamed down when the railroad brought people and their economic interests. More generally, lawlessness ends when persistent infrastructure brings visibility and an economy produces something people are incentivized to protect.

Infrastructure will beget infrastructure. Autonomous vehicles will create demand for energy nodes, which will enable longer missions. Longer missions will generate more data, which will create demand for communications networks. Each layer will make the next layer viable and necessary at greater scale.

Commercial activity will overtake military spend. Defense is always the early customer for frontier infrastructure. Before the Gold Rush transformed it into a commercial hub, San Francisco began as a Catholic mission and a military fort, the Presidio; today the Bay Area generates $1.4 trillion in GDP and would rank as the world’s sixteenth-largest economy if it were a country. Just as military forts preceded commercial towns in the West, military satellites dotted the skies before telecommunications satellites. The military cares more about capabilities than cost and can therefore bootstrap markets before the economics pencil out. But as always, the commercial market will dwarf defense in time via cable monitoring, offshore energy, fisheries, shipping, and a number of unknown but inevitable use cases that always emerge when costs come down.

New industries will emerge that nobody can predict today. I’m cheating a bit here, because this happens every time. The most valuable businesses that emerge from frontier industrialization are often the ones that people figure out once they have time to get their hands dirty with new capabilities. It’s not a coincidence that a disproportionate number of the richest people who have ever lived (Vanderbilt, Carnegie, Rockefeller, Musk) built the access infrastructure at exactly the right moment and then created new markets around the new capabilities they brought into the world.

That said, we can imagine some specific opportunities, including but not limited to: managed fisheries at scale; critical mineral extraction with proper environmental monitoring; persistent maritime domain awareness sold as a subscription; autonomous shipping and smarter logistics; ocean energy at industrial scale; offshore compute powered by ocean energy and cooled by seawater; an undersea acoustic and optical communications network (the subsea equivalent of Starlink), connecting every vehicle and sensor in the water; a persistent ocean observation layer (the subsea equivalent of Planet Labs) continuously mapping the ocean in three dimensions; systematic scientific exploration of the 99.9% of the deep seafloor that nobody has ever studied.

There are an estimated three million shipwrecks on the ocean floor, most of them never found, collectively carrying billions of dollars in treasure and untold secrets about the human story. And we may finally realize our parents’ and grandparents’ dreams of permanent ocean habitats (seasteading, seriously: I think it could actually work in the next decade with the right supporting logistics, communications, and energy infrastructure). A new Atlantis is at hand.

The scale of what opens up when you collapse the cost of persistent ocean presence is difficult to overstate. The ocean economy today is built on transit, extraction, and coastal adjacency, and it’s already worth trillions. Imagine what happens when you add persistent infrastructure, domain-wide monitoring, in-water energy, autonomous logistics, undersea communications, and a continuous observation layer covering 70% of the planet. The new economy will dwarf the old one. We just can’t see most of it yet, and that’s exactly what you’d expect at this stage.

We return to the question at the heart of my mission. What does industrialization mean for the health of the ocean itself?

There is a widespread assumption that economic activity in the ocean and ecological health are in tension and that industrialization means exploitation. I believe the opposite is closer to the truth.

Counterintuitively, with all of this economic activity, we will get conservation that actually works, practically for free.

The technologies that industrialize the ocean are the same technologies that make real conservation possible for the first time.

Consider the state of ocean conservation today. Marine protected areas cover roughly 8% of the ocean on paper. In practice, many of them are “paper parks,” with no enforcement and no monitoring. We protect the ocean by drawing lines on maps and hoping people respect them. I know this because I have worked in marine protected areas on both coasts of Australia and the US. I can tell you that it doesn’t work for the same reason that protecting the Amazon didn’t work until Brazil launched real-time deforestation alerts in 2004, after which Amazon deforestation fell 70%. Visibility is what made the difference. You can’t conserve what you can’t see.

Persistent ocean infrastructure gives you that visibility, as well as the ability to intervene. We won’t know what’s possible until we experiment with the right equipment and measurement.

Consider the story of Russ George.

In 2012, George, a Californian entrepreneur, tried one of the boldest and most controversial experiments in the history of ocean science. He decided to implement the work theorized by the controversial oceanographer John Martin, the father of an idea called “ocean iron fertilization”. A few decades earlier, Martin stood up at a scientific conference and declared: “Give me half a tanker of iron, and I’ll give you an Ice Age.”

Martin’s hypothesis was that iron fertilizes ocean plankton, which multiplies and absorbs enormous quantities of CO₂. This then feeds the base of the marine food chain, which, if done at scale, might be able to alter the planet’s climate and restore fish stocks. He died before he could test it properly.

In 2012, Russ George picked up his work and convinced a First Nations village on Haida Gwaii, a remote archipelago off the coast of British Columbia, to invest $2.5 million in the experiment. He loaded 100 tonnes of iron sulfate onto a fishing boat and dumped it into the Pacific about 200 miles offshore.

And by George, he did it. NASA satellites confirmed a plankton bloom covering 35,000 square kilometers — an area so large that it was visible from space. The following year, the pink salmon run returned at 226 million fish, the largest run in Canadian history.

The scientific establishment lost its mind. Researchers disputed the causal link. Environment Canada executed search warrants on his office. The Canadian government investigated George for violating international dumping conventions. The UN imposed a moratorium on ocean fertilization.

None of his opponents argued that the idea was wrong, nor even that it didn’t work. The problem was that nobody could verify what had actually happened. There was no persistent monitoring to measure the bloom’s full effects, no way to track downstream consequences across time and space, and no ability to adjust the intervention as conditions evolved.

George had a hypothesis and a budget, but he was intervening blind. One rogue entrepreneur with a fishing boat had produced what looked like one of the most successful ecological interventions in history, and nobody could prove it, so nobody has reproduced it.

Now imagine that same experiment in a fully instrumented ocean. Autonomous vehicles monitoring the bloom in real time across its full extent. Sensors tracking nutrient levels, oxygen, pH, and biodiversity indicators continuously. The ability to see unintended consequences developing and respond before they cascade. Real-time data flowing to scientists and regulators who can assess the intervention as it unfolds, not years later from fragmentary satellite data. Iron fertilization is just one example. Targeted reef restoration, invasive species removal, pollution response, managed rewilding: these are the kinds of active stewardship that become possible when you can actually see what’s happening in the ocean and respond to it. The problem was never the idea of active ocean management. The problem was attempting it without the infrastructure to close the feedback loop.

The same sense-intervene-measure-adjust feedback loop that George’s experiment lacked is exactly what persistent infrastructure provides.

The dual-use nature of ocean infrastructure could make all of this work. The same vehicles monitoring subsea cables will collect environmental data. The same platforms enabling offshore energy will measure ocean chemistry. The same communications networks connecting autonomous fleets will transmit scientific data. It is inefficient to build one system for commerce and another for conservation when you can run both on the same system.

Economic growth and biodiversity health can and should be genuinely symbiotic. The fishing industry benefits from healthy stocks, which require monitoring. The offshore energy industry benefits from accurate ocean data. The insurance industry benefits from better climate models. Every commercial use case generates data that improves our understanding of the ocean, and better understanding leads to better stewardship. We still need to respect the ocean as a living system, and there will be hard decisions about where and how to operate, but we need to have the ability to decide. The virtuous cycle only works if the infrastructure is in the water.

I am not naive about this. There are real risks. Poorly managed extraction can damage fragile ecosystems. Increased activity in previously untouched areas brings the possibility of contamination. Geopolitical competition over ocean resources could intensify rather than stabilize. The fact that international waters have no FAA equivalent is an opportunity, but it is also a governance gap that will need to be filled as activity scales. Getting this right requires exactly the kind of persistent monitoring and real-time visibility that ocean infrastructure provides. The best argument for building the infrastructure is that without it, the ocean’s resources will continue to be exploited blindly and unsustainably by those who can already afford to be there.

When you fix the cost of doing persistent work in the ocean, you fix everything downstream of it. Conservation, commerce, defense, science, and discovery are all bottlenecked by the same constraint. If we remove it, we can unblock all of them.

We Must Return to the Great Blue Frontier

Berry Cannon died at 610 feet, breathing helium in the dark, because somebody forgot to fill a scrubber. That was 1969. That was the best we could do back then.

We can do much better today. The technology exists to build persistent, autonomous infrastructure across the ocean without sending human beings into the abyss. But the point of this essay is not that we can finally do it safely. The point is convincing you that we have to do it at all.

We have the means now to conquer the Great Blue Frontier and do what our grandparents dreamed of. Frontiers matter. Conquering them matters. Not just economically, though the economic case is extraordinary, but for what they do to us and more broadly, the soul of civilization. Their pursuit lights a fire in the heart of men that leads to civilisational greatness.

Every great era of advancement has corresponded to an active frontier. The Renaissance had the Age of Exploration. The twentieth century had aviation, then space. When there is somewhere new to go, the best minds build toward it. When there isn’t, that energy turns inward. It optimizes. It litigates. It fights over what already exists. I think we’ve all felt that. The creeping sense that the great adventures are behind us. That the best we can do is make what we have a little more efficient.

Civilization is a lot like sharks in this regard. Many species of shark (e.g., Great Whites, Makos, Whale sharks) die when they stop swimming. They need the forward motion to oxygenate their gills. Stop, and they suffocate. Humans are the same. Without forward motion, we stagnate, and stagnation is a kind of death.

The ocean is the answer to this stagnation. It has been there the whole time, covering 70% of the planet, largely unknown, barely touched, immensely rich. We mapped its surface and called it quits. We drilled through it and called it industry. We fished from it and called it the economy. But we never truly entered it, and we certainly never built the infrastructure to stay, steward, and leverage its full potential

The last great frontier on Earth isn’t behind us. It’s below us.

I grew up jumping off the pier at Garrettstown, pulling crabs out of rock pools, heading out in the boat with my dad to catch mackerel. I loved the ocean before I understood it, before I could have told you anything about technology curves or governing metrics or the cost of persistent presence. I just knew there was something down there worth knowing. I still believe that. And now we can finally go find out.

Come build it with us. There is no greater adventure than settling a frontier.

Will O’Brien is the co-founder and President of Ulysses Maritime Technologies, the ocean company, building the fleets of autonomous underwater and surface vehicles to solve the most critical tasks in the ocean.

Thanks to Will for sharing his knowledge, to Badal for my favorite cover art yet, and to all of the oceanic explorers buried beneath the waves.

That’s all for today. We’ll be back in your inbox tomorrow with another Weekly Dose.

Thanks for reading,

Packy

Hi friends 👋,

Happy Friday and welcome back to our 189th Weekly Dose of Optimism.

There are a couple of big themes this week: energy, and non-invasive, non-molecular control of the human body and brain. If we’d published this week’s Dose back in, say, 1995, I worry that a few readers would drop dead from shock. It’s hard to recognize you’re in a sci-fi novel when you’re living through it, but if you zoom out a little bit, it sure looks like we are.

I’d highly recommend the Extra Doses this week, for Friedberg contextualizing everything happening and why it’s good, and for one of the craziest stories we’ve covered in Science Breakthroughs on gene control in mice using electromagnetic fields. Plus, Science Breakthroughs now have one-sentence plain English summaries thanks to a request from Chris B.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Create

Funnily enough, I’m writing this as I sip my morning Creatine + Electrolytes. It’s something of a morning ritual in our house. The kids wake up and come downstairs, Maya asks if we can make creatine, we open two packets, she pours them in, Maya and Dev each take a small sip, and then I down the rest. Dan is making me do a Hyrox with him, so C+E is a lifesaver.

I’m apparently not alone. Since Dan and Sienna launched Creatine + Electrolytes two weeks ago, the product is already run-rating $15 million. And no wonder.

Each packet contains 5g creatine monohydrate, 800mg electrolytes, and 1000mg taurine. I take two in the morning, and one or two later in the day when I work out or if I’m tired. These little packets, mixed into a glass of water, support hydration, improve recovery, increase energy, and boost cognition.

They have all of the well-studied benefits of creatine, plus the hydration, muscle function, recovery, nerve signaling, heart rhythm, and pH balance benefits of electrolytes. Before they put it all in a little packet, I was dumping salt in my creatine water. This is better, and it tastes so good that my kids demand sips.

So if you want a great-tasting extra kick for your muscles and your brain, and you want to support long-time Dose writer and my brother Dan, get 30% off C+E here:

Get 30% Off Create C+E

(1) Quaise Energy Introduces World’s First Superhot Geothermal Plant

Beneath our feet sits essentially infinite, always-on, carbon-free power. Estimates are that there 63 terawatts, or roughly eight times all the electricity humanity currently generates, would become accessible by tapping just 1% of the world’s superhot rock resources. It’s just really far down there. The hottest, densest heat sits two to twelve miles down, past the point where conventional drill bits work.

This week, MIT spinout Quaise Energy announced Project Obsidian: the world’s first commercial superhot geothermal power plant, being built in Central Oregon near the Newberry Volcano. Phase I will be 50 MW, Phase II will scale to 250 MW, and Phase III will be more than a gigawatt. Quaise is targeting commercial operations in 2030.

The reason this can exist now is a new way to drill. Quaise uses a gyrotron, originally developed for fusion research at MIT’s Plasma Science and Fusion Center, to produce millimeter-wave energy that ablates rock by vaporizing it with no mechanical contact. Last year, they drilled through 100+ meters of granite in Central Texas in the first field demonstration of the technology. This year, they’re targeting a kilometer, then eventually, 10-12 miles. At full depth, a single superhot well would produce 5-10x more power than a conventional geothermal well.

Both the superhot rock and the drill that can reach it are what make this different from Fervo, a company we’ve covered previously in the Dose, and the other EGS players working closer to the surface. Conventional EGS taps warm rock. Quaise is going for the really hot 300–500°C rocks, which provide enough pressure to produce supercritical water, which makes gigawatt-per-well plausible.

We have been waiting for this. When Julia and I talked to Eli Dourado about geothermal on Age of Miracles, he said the reason we couldn’t just count on geothermal for all of our electricity needs was heat. “Geothermal doesn’t get super hot,” Eli said.

If Quaise pulls this off, that may no longer be true. Drill baby drill.

(2) Panthalassa Makes Wave-Powered Nodes for Offshore AI DCs

and Garth Sheldon-Coulson on

A year or so ago, I had coffee with Lowercarbon Capital’s Ryan Orbuch, and I asked him what the coolest companies in his portfolio were. The first one he brought up was Panthalassa, which I remember him describing as building massive structures in novel geometries to pull energy from the ocean’s motion (turns out both the size of the boat and the motion of the ocean matter).

Panthalassa, based out of Portland of all places, has been working on an 80-meter-tall, 20-meter-wide, lollipop-shaped self-driving floating power plant that lives on the open ocean miles and miles offshore and generates electricity from waves, which it will first use, of course, to power GPUs. The lollipops will have just one moving part. Waves force water through internal channels, the channels pressurize the water, the pressurized water spins a turbine, and the turbine spins a generator.

CEO Garth Sheldon-Coulson told Ashlee Vance that they chose offshore waves because it’s one of three potential tens-of-terawatt-scale energy sources (the other two are solar and nuclear (fission and fusion), and there’s a potential fourth in the next story). When he and co-founder Brian Moffat dug in, they realized it also had advantages over other renewable sources: capacity factor up to 90%, against 30-40% for offshore wind and 25% for solar and a target cost around $1,500 per kilowatt. It can also self-propel to wherever the wave resource is best at a given time at 1-1.5 knots. It all sounds insane, but results from sea trials of their Ocean-2 prototype in Puget Sound matched simulation almost exactly.

If this works, we’re going to need to figure out a lot more things to do with all our power.

(3) Vital Lyfe Opens Up Access for Pre-Orders

One of which may be to desalinate the very water in which Panthalassa swims.

Look, I love putting Creatine + Electrolytes in my water more than the next guy, but there are plenty of times when you want to get the salt (and other stuff) out of your water to make it safe to drink. For those times, there is now Vital Lyfe’s Access.

Jon Criss (13 years at SpaceX, Lead Integration and Test Engineer on Dragon, Product Manager on Starlink) and Andrew Harner (9 years at SpaceX, Stanford aero) co-founded Vital Lyfe to build what they call a personal water-making system. This week, pre-orders went live for Access: a 25-pound box that turns seawater into drinking water. Put a hose in the ocean, press start, and collect six gallons of clean water per hour. Freshwater sources get you 12-13 gallons.

Access is $749, with an $8 deposit to reserve, and deliveries begin later this year.

Desalination has always worked. Israel runs its whole country on it. The problem is that it has required municipal-scale infrastructure like billion-dollar plants, dedicated power, and pipelines, which means that the 2+ billion people who lack reliable access to clean water can't get it, because it depends on the kind of infrastructure their governments haven't built and probably won't.

Vital Lyfe's bet is the same bet that made Starlink work: if centralized infrastructure can't reach everyone, decentralize the infrastructure. Make the unit small enough that an individual, an NGO, a fishing village, a disaster response team can just… have one. In March, they field-tested Access in Tierra Bomba, Colombia, with local NGO Amigos Del Mar. They opened a 37,000-square-foot manufacturing facility in Torrance the same month. Criss told press the production line is designed to manufacture more desalination units in a single month than currently exist worldwide.

Drink up.

(4) This Beanie Is Designed to Read Your Thoughts

Humans have been wearing hats for at least ~30,000 years. We’ve been thinking even longer than that. What if, Sabi asks, we could just think while wearing a hat that reads our brain, and use that to interact with machines?

Yesterday, Sabi, backed by Khosla, Accel, Initialized, and OpenAI VP of Science Kevin Weil, came out of stealth with a BCI that decodes the words you think into text on a screen, but instead of needing brain surgery, the BCI comes in a hat.

The device uses 70,000 to 100,000 EEG sensors, versus a dozen to a few hundred for most consumer EEGs, all in a baseball cap or beanie. The bet is that a ton of sensors can compensate for the skull’s signal dampening, the problem that most BCI companies compensate for by going under the skull. “If you're going to have a billion people use BCI for access to their computers every day,” Vinod Khosla said in a WIRED interview, “it can't be invasive.”

Sabi is also building a Brain Foundation Model to decode the brain’s signals and turn it into speech. They’re targeting 30 words per minute, but the plan is to improve the speed as the hats learn about their wearers’ brains.

Sabi expects to begin shipping at the end of the year, so now is the time to start meditating and controlling your thoughts before your hat starts telling everyone around you what you really think about them.

(4b) Prophetic Successfully Demonstrates Increased Lucidity in Dreams

Speaking of non-invasive brain devices… we’ve written about Prophetic, a company that plans to use Transcranial Ultrasound to induce lucid dreams, a few times here in the Dose, for the simple reason that it’s one of the coolest companies imaginable. It is also an expression of a belief I have that as the machines get more capable, we will see a wave of technologies built to unlock the superpowers latent in our human bodies and minds.

Good news for the humans: Prophetic has successfully demonstrated the ability to increase lucidity in dreams. “By sending safe ultrasonic energy through the forehead and into the prefrontal cortex,” the company tweeted, “we are able to activate the Central Executive Network aka Frontoparietal Network.” Test subjects reported greater dream recall, dream vividness and continuousness, clearer dream thinking, and the ability to make deliberate choices that changed what happened inside of the dreams.

Lucid dreams are wild, with benefits (studied to varying degrees) around nightmare reduction (including PTSD-related ones), motor skill practice (you can practice throwing darts in a lucid dream and get better at actually throwing darts), emotional processing and fear exposure, and creative problem solving. Nikola Tesla wrote about his deliberate conscious dream-journeys in his excellent and short autobiography My Inventions, writing “Every night, when alone, I would start on my journeys – see new places, cities, and countries; live there, meet people and make friendships.” Christopher Nolan created Inception on concepts from the lucid dreaming literature, and said in interviews that he drew on his own dream experiences for the film.

Aside from the practical benefits, lucid dreaming just seems magical, like VR that the brain you always have with you can generate. In lucid dreams, people can fly, interact with dream characters, and generate worlds that defy the laws of physics. The world needs more magic.

(5) Salk Institute Gets $41.3M from ARPA-H to Advance Sonogenetics

Ultrasound is making a lot of noise this week. Salk Institute neuroscientist Sreekanth “Shrek” Chalasani received up to $41.3 million from ARPA-H to turn the technology he invented, sonogenetics, into a clinical therapy.

Where Prophetic is using ultrasound to stimulate brain regions we already know respond to acoustic energy, sonogenetics flips the approach: you engineer specific cells to express an ultrasound-sensitive protein, then use a focused ultrasound pulse to turn just those cells on or off. It’s like optogenetics (using light to control neurons), except ultrasound goes through skin and bone without needing fiber-optic implants. It’s also more precise than a drug, because only the tagged cells respond.

Chalasani found the first ultrasound-sensitive protein in C. elegans in 2015, coined the term "sonogenetics," and spent the past decade building the field on seed funding from the Jacobs family at Salk. The ARPA-H award pulls together a team built for this kind of translation: Nobel laureate Ardem Patapoutian at Scripps engineering the next generation of ultrasound-sensitive proteins, Aravind Asokan at Duke working on delivery vectors (Go Duke), Xuanhe Zhao at MIT on wearable ultrasound hardware, plus collaborators at UC San Diego, St. Boniface/Manitoba, and the California Medical Innovations Institute. Salk spinout SonoNeu (co-founded with Venkat Reddy at General Inception) will manage the path to FDA.

Their first target is peripheral neuropathies, including diabetic neuropathy. Longer-term, they want to build a platform for on-demand, location-specific control of cell activity in both the peripheral and central nervous system.

One of the things I find most exciting about biology today is how many of the big frontier ideas are about programming living systems with something other than small molecules: Michael Levin on bioelectrics, Prophetic on ultrasound at the circuit level, Chalasani on ultrasound at the cell level. Welcome to the Body’s Electric Age.

EXTRA DOSES: Science Breakthroughs, Friedberg on Modern Wisdom

Read more

Hi friends 👋,

Happy Friday and welcome to our 188th Weekly Dose of Optimism!

I keep saying this, and then it gets cold again, but I think this is the one: it’s a beautiful morning here in New York City, spring is springing, and our fellow humans keep pumping out optimism-worthy work.

This week, for example, we have new funding for AI-assisted Alzheimer’s research and, in Science Breakthroughs, a peptide that degrades the amyloid-β linked with the disease. That, and much much more, in this week’s Dose.

Let’s get to it.

Today’s Not Boring is brought to you by… Silicon Valley Bank

There is a lot of noise out there about how AI is impacting companies. Of course, big AI lab CEOs are predicting massive job disruption and productivity increases. But to find out what’s really going on, ask the people who control where the dollars flow: the CFOs.

SVB surveyed 200+ startup finance leaders for its new State of the VC-Backed CFO report, and a few things jumped out:

• AI adoption is now the #1 issue for startups, and 63% of CFOs rank it top-two

• Companies are doubling AI spend this year (to ~$50K median)

• Over half of CFOs surveyed are already seeing real ROI on AI spend

• The biggest impact to staffing isn’t layoffs, it’s hiring fewer junior hires

In short, startups are getting leaner, faster, and more experimental. If you want to understand where company-building is going next from the people who control the spend, grab your free copy today.

Read SVB’s State of the VC-Backed CFO

(1) OpenAI Foundation Commits $100M to Cure Alzheimer’s

My grandmother had Alzheimer’s. One of my biggest fears in life is that me and the people I love will get it; it’s one of the reasons I started writing, to help me remember. Unfortunately, since my grandmother had it, we’ve made no progress on defeating Alzheimer’s. Measured by age-standardized mortality rate, while we’ve made strides against cardiovascular disease, cancer, and infectious diseases, Alzeheimer’s has only gotten worse. It is a complex disease with hundreds of environmental and genetic risk factors interacting across cell types over decades, and as such, treating it has been a sad game of whack-a-mole.

Maybe AI can help. This week, the OpenAI Foundation, the nonprofit parent of OpenAI, now sitting on a $25 billion life sciences and resilience commitment, announced it’s putting $100M+ into six research institutions to throw modern AI at the problem. Jacob Trefethen, who ran half a billion dollars of science grantmaking at Coefficient Giving before joining OpenAI, is leading the effort. The grants span AI-assisted drug design, biomarker discovery, mapping disease pathways, and personalizing treatment.

The most interesting grantee is Arc Institute, whose progress we’ve covered many times in the dose, which announced a parallel partnership focused on what co-founder Patrick Hsu calls an “AI lab-in-the-loop” approach. The idea is to systematically perturb human brain organoids guided by patient data, measure what happens, feed the results back into AI models, and iteratively build a causal map of the disease, the actual chain of cause and effect from genetics to protein misfolding to synaptic collapse. Arc executive director Silvana Konermann describes the goal as finding “perturbations that can click and drag a cell from a diseased state back into a healthy one.” The other grantees include David Baker’s Institute for Protein Design and EvE Bio.

NIH funds Alzheimer’s, and has for decades, and we are where we are. What Arc and the OpenAI Foundation are doing is what private capital and motivated foundations can do that institutional science usually can’t: pick a hard problem, fund a full-stack experimental-and-AI engine, and run the loop fast enough that we might actually get somewhere by the time it matters to my family.

I would very much like to forget about Alzheimer’s once and for all.

(2) Information to Atoms

Substrate x Google DeepMind

One big lab I have been consistently impressed with is Google DeepMind, and those wacky geniuses are at it again. This week, the American semiconductor foundry startup Substrate, whose launched we kicked off with in Dose #168, shared that GDM’s AlphaEvolve rewrote a chunk of its lithography software and made it nearly seven times faster while cutting compute costs by 97%.

Substrate is the James Proud-founded startup that came out of stealth last October with $100M and a crazily ambitious mission: build a U.S. foundry that uses X-ray lithography to compete with ASML's EUV machines by patterning features tens of atoms wide, on American soil, at lower cost than the Dutch monopoly. AlphaEvolve is DeepMind’s evolutionary coding agent, the latest entry in the Alpha lineage that runs from AlphaGo to AlphaGoZero to AlphaZero to MuZero to AlphaFold to AlphaChip.

This week, Substrate pointed AlphaEvolve at its computational lithography stack, the software that simulates how trillions of photons interact with photoresist to print a chip layer. In just a few weeks, AlphaEvolve explored thousands of algorithmic variations and landed on lossless compression tricks and lower-precision representations that cut memory by 74%, sped up runtime 6.8x, and dropped the Google Cloud TPU bill by 97%. Now, Substrate’s tool can print metal-one layers (the hardest, most defect-prone layer in a chip) at a 24nm pitch in a single exposure, with bidirectional 2D patterns and sharp 90-degree corners. That’s 2nm-node territory, the bleeding edge. Without the speedup, the same patterns would have required multi-patterning: two or three exposures, which means more defects, more cost, and more time.

This is fun because: a) we are rooting for American ASML and b) it’s a combination of the stuff we’ve talked about in the last couple of co-written essays: World Models talked about the evolution of the Alpha line of models, which learn causality instead of rules, and Electromagnetism Secretly Runs the World talked about using AI to design better chips. It’s also an example of America’s best hope in the electric competition with China: using new technologies to find new, better, and cheaper ways to build the things we don’t build today. I suspect we’ll be hearing more from Substrate - it would be insane to compete with ASML at their own game, so they’re creating a new one.

(2a) Arena Physica Releases Heaviside

Speaking of EM modeling… A couple of weeks ago, in Electromagnetism Secretly Runs the World, Arena Physica CEO Pratap Ranade and I mentioned the new models and blog posts they would be releasing soon. Soon has come. Its new model, Heaviside, is its first EM foundation model, and you can play with it in the RF Studio.

(3) Syncere Unveils Lume Lamp That Folds Your Laundry

I have spent a lot of time thinking about robotics recently, and I’ve got to admit, I did not think of a lamp that folds your laundry. This is the future that Beauty and the Beast and The Brave Little Toaster promised.

Lume is a pair of six-foot bedside floor lamps with hidden six-axis robotic arms, cameras, and on-device AI processors. Toss your clean laundry on the bed, the lamps quietly extend their arms and fold it in under two hours for a full load. You don’t have to tell it to do anything, it just watches the bed and figures out when to get to work. (There’s a Lume sitting in the factory somewhere just praying Bryan Johnson orders it.)

The graveyard of laundry-folding robots is packed. Foldimate made you feed clothes in one at a time and quickly… folded. Laundroid was a $16,000 closet-sized monster that bankrupted its parent company in 2019. Laundry is so hard that every robotics company makes demos of their bots figuring out how to do it in a controlled environment, and people go wild, because folding laundry is decidedly a thing we want the robots to do for us.

The why now? is the perception stack. Vision-language models (VLMs) are now good enough that a small startup can train a thing to look at a balled-up t-shirt on a comforter, figure out what it is, grab it, and fold it.

It’s also a constrained problem. It doesn’t promise to do everything; it gets paid to do one thing well, and learn. Maybe it, like Standard Bots, will be able to take small steps across the spectrum. And you gotta buy a lamp anyway.

I want to wait to see the reviews before I drop $2,499 on two lamps, but if you want to be an early adopter, be our guest.

(4) Humans Travel Farther From Earth Than Ever Before

Jared Isaacman for NASA

Last week, we covered the launch of the Artemis II mission, which planned to take North Americans back to the Moon for the first time in more than half a century, and farther from the home planet than we’ve ever traveled (that we know of). As the crew rounded the Moon on Monday, NASA Administrator Jared Isaacman tweeted:

On the far side of the Moon, 252,756 miles away, Reid, Victor, Christina, and Jeremy have now traveled farther from Earth than any humans in history and now begin their journey home. Before they left, they said they hoped this mission would be forgotten, but it will be remembered as the moment people started to believe that America can once again do the near-impossible and change the world.

It’s rare for us to do back-to-back coverage of the same story, but unprecedented accomplishments call for unprecedented measures. Instead of writing about it, I’d just ask you to watch Astronaut Victor Glover’s Easter message back home:

(5) Primer Launches New Brand with The Orator

Primer is one of the not boring capital portfolio companies I’m most proud to be involved with. I’ve written about them here and talked to the CEO Ryan Delk on moving from software into running schools here. They take kids seriously, and exist to give more kids access to excellent education.

This week, they released a video narrated by one of its students and featuring others that highlights both its approach to education, and an optimistic vision for how we can all meet the future: “The next 200 years will look nothing like the last. We will face problems we can’t yet imagine, and possibilities we can barely dream of. We won’t just need answers. We’ll need resilience, flexibility, courage, mastery.”

Amen.

EXTRA DOSE: Science Breakthroughs, Hermeus, Pace Fellowship

Read more

Hey friends 👋,

Happy Friday and Happy Easter Weekend!

We’re going back to the Moon. OpenAI bought TBPN for a lot of money. Somos is going to blanket the globe in fast internet. And a whole lot more good news where that came from.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Guru

Your company has invested in AI tools, which is a great first step. But your AI is only as good as the knowledge behind it. And for most companies, that knowledge is scattered, outdated, and nobody’s job to fix. So your AI gives confidently and sneakily wrong answers to all of your employees.

Guru is the governed knowledge layer that fixes this at the foundation. Companies like Spotify and Brex use Guru to structure their company knowledge, keep it accurate automatically, and make it available to every AI tool and every person, all from one trusted source.

With Guru, when an expert corrects something once, it’s right everywhere you work, in every AI tool, workflow, and answer.

No more garbage in, garbage out. Just cited, verified, always-current knowledge your AI can actually rely on.

Try Guru Today

(1) NASA’s Artemis II Crew Launches to the Moon

NASA

For the first time in 53 years, 5 months, and 20 days, human beings are on their way to the Moon.

Last night at 6:35 p.m. EDT, NASA’s Space Launch System (SLS), standing 322 feet tall, weighing 5.7 million pounds, and generating 8.8 million pounds of thrust (17% more than the Saturn V), lifted off from Pad 39B at Kennedy Space Center carrying four astronauts aboard the Orion spacecraft (which the crew named Integrity). Commander Reid Wiseman, Pilot Victor Glover, and Mission Specialists Christina Koch and Jeremy Hansen are now on an approximately 10-day trip around the Moon and back. No one has done this since Apollo 17 splashed down in December 1972.

Integrity will spend about a day in a high Earth orbit stretching 46,000 miles out, the crew will manually fly the capsule to test its handling, and then a six-minute translunar injection burn will send them on a free-return trajectory using the Moon’s gravity to slingshot them around the far side and back home. On Monday, April 6, the four of them will become the first humans to lay eyes on parts of the lunar far side, reaching a record distance of 252,000 miles from Earth. They’ll splash down in the Pacific on April 10.

Is getting to the Moon more expensive than it should be? Hell yeah, roughly $4.2 billion per SLS flight by the Inspector General’s math. Did it take way longer than planned? Also hell yeah. But four people are about to travel further from our home planet than any human has traveled before, in the first of a series of missions designed to get us back to the Moon to stay.

I can’t really sum up how I feel any better than this kid did:

(2) OpenAI Acquires TBPN

Katie Deighton for WSJ

Speaking of going to the f*cking moon… OpenAI, which just closed a $122 billion round at an $852 billion post-money valuation, has spent some of its treasure to acquire tech media’s crown jewel, TBPN. The FT reported that the deal is in the “low hundreds of millions of dollars.”

Huge congrats to Jordi, John, Dylan, Ben, and the whole TBPN crew.

This is an incredibly smart acquisition for OpenAI. I’ve been saying that the frontier AI labs have completely botched their narrative, and this is about as good of a way to get to work fixing it as I could imagine.

A lot of the commentary so far has been about OpenAI acquiring the media property, which will certainly be helpful, but the big get is Jordi & John. They’re simultaneously highly generative creative geniuses and normal guys. I expect that we’ll see their impact on the way the company communicates to non-San Franciscans pretty quickly, and I hope they’ll have a hand in shaping where AI goes more broadly.

In the meantime, it’s just great to see the good guys get a W.

(3) Google DeepMind Releases Gemma 4 Open Source Model

Completely unrelatedly, I’ve told you guys how awesome I think Google DeepMind is, right? Been saying it for years. What a great lab. Could obviously use a little help telling their story, but wow. What a family of models. AlphaFold! Gemini! AlphaProof! AlphaGo! AlphaZero! Nano Banana! Genie! SIMA! DolphinGemma! I could go on and on.

Yesterday, the good folks at GDM just released a family of AI models that can see images, watch video, understand speech, reason through competition-level math, and write code, all open source. You can download them for free, modify them, and ship them in a commercial product under the Apache 2.0 license.

GDM released Gemma 4 in four sizes: a 31B dense model, a 26B mixture-of-experts model, and two edge models (E2B and E4B) designed to run on phones and laptops. None of them is quite up to par with the best frontier models, including GDM’s Gemini, but since they can run locally on your hardware, it means you get the tokens they produce basically for free.

Yesterday, in Bad Analogies, I wrote that OpenAI and Anthropic were burning a ton of money to compete in an incredibly competitive field, and that Google DeepMind was Zeus in this battle of the gods. Gemma is another lightning bolt. The more things people can do with free tokens, the less they’ll spend on tokens, even if the most advanced capabilities will always come at a price. Of course, closed model customers are paying for more than tokens: reliability, ease-of-use, etc… but it’s fun to see the competition drive down prices for consumers.

Jokes aside, I continue to be very impressed by GDM. The lab came up more than any other in my research for both World Models and Many Small Steps for Robots; they’ve done more to push the frontier in more directions than any lab in the world. With Gemma 4, Demis the Menace strikes again.

(4) Solar cells just did the “impossible” with this 130% breakthrough

Kyushu University via Science Daily

Three weeks back, we led off the Dose with a story on Swift Solar buying a gigawatt-scale solar factory to produce 30-40% efficiency silicon-perovskite tandem solar cells.

This week, researchers at Kyushu University in Japan announced cells with a seemingly impossible 130% efficiency.

There’s a number in solar energy called the Shockley-Queisser limit. It says that a single-junction silicon solar cell can convert, at most, about 33% of the sunlight hitting it into electricity. Silicon-perovskite tandems can up that to 45%. In either case, the rest, especially the high-energy photons, gets wasted as heat. Shockley-Queisser been the ceiling since 1961.

Until now! A team led by Associate Professor Yoichi Sasaki at Kyushu University in Japan, collaborating with Katja Heinze’s group at Johannes Gutenberg University Mainz in Germany, just published a paper in the Journal of the American Chemical Society showing that they paired tetracene-based molecules with a molybdenum-based metal complex (a “spin-flip” emitter) and achieved a quantum yield of roughly 130%. That means for every photon absorbed, the system produced 1.3 energy carriers. More out than in, which seems like it shouldn’t happen.

The trick is a process called singlet fission. When a high-energy photon hits the tetracene, instead of creating one excited electron and dumping the rest as heat, the molecule splits that energy into two lower-energy “triplet” excitons, a little bit like nuclear fission (don’t yell at me real physicists). The problem has always been that a competing process called Förster resonance energy transfer tends to steal the energy back before fission can finish.

Sasaki’s team solved this with the molybdenum complex, which selectively accepts only the multiplied triplet excitons by flipping the spin of its electrons during near-infrared emission. If molybdenum sounds familiar, it’s because it’s the hard-to-pronounce mineral Earth AI found in its first discovery, and because it’s come up a couple of times in recent American Alchemy episodes.

There is an asterisk, for now: the process was demonstrated in solution, not in a working solar panel. It’s a proof of concept. The team’s next step is integrating the tetracene and molybdenum materials together in a solid-state system.

But the result proved that the 100% quantum yield barrier isn’t a physical law, which means the engineers can get to work sorting out the details. If singlet fission can be made to work in commercial cells, we’re going to start putting a lot more photons to good use.

(5) Somos Internet Raises $40M Series B

Alex Konrad for Upstarts Media

In November, I released my longest company Deep Dive ever: Cable Caballero, the story of Forrest Heath III and the vertically integrated internet company he and his band of pirates are building down in Medellín, Colombia.

Explaining how he built what is now the best-performing internet in the region, and maybe the world, with its own fiber network, switches, routers, and ISP, Forrest said:

It’s been this never-ending game of doing something janky, getting credibility, doing crazier stuff, getting more resources, getting smarter people so that we can fix the things that were messed up in the janky past iteration, then gaining credibility to get more resources to get cooler people to do crazier stuff. It’s like this self-sustaining fission process.

Now, Somos has a lot more resources to hire more smart people and do even crazier stuff! This week, the company announced that they raised a $40 million Series B co-led by Bracket Capital and Ribbit Capital. Following on to not boring capital’s original investment was a no-brainer.

Somos is growing fast in Colombia, and it announced that it’s getting ready to expand into Mexico. On TBPN, Forrest even hinted at Somos Caracas…

I think this company is going to build a ton of better, faster, cheaper infrastructure in the coming decades, and they have both the fuel and the backers to do it. If you’ve been looking for a sign to put your stuff in a bag and move to Colombia, this is it.

Vamos Somos!

EXTRA DOSE: Science Breakthroughs, Electric Delivery, Bangalter

Read more

Subscribe now

Hi friends 👋,

Happy Thursday! I wasn’t planning on sending an essay this week. I wanted to give you a little time to breathe after World Models and Electromagnetism.

But then I saw someone make a bad point on the internet, and it was a particular bad point that has always bugged me, so I whipped up a short and sweet rebuttal. In short: not every company that burns a lot of money is the next Amazon. (a bit of a strawman, I admit, except this analogy keeps popping up!)

Let’s get to it.

Today’s Not Boring is brought to you by… Framer

Framer gives designers superpowers.

Framer is the design-first, no-code website builder that lets anyone ship a production-ready site in minutes. Whether you’re starting with a template or a blank canvas, Framer gives you total creative control with no coding required. Add animations, localize with one click, and collaborate in real-time with your whole team. You can even A/B test and track clicks with built-in analytics.

Launch for free at Framer dot com. Use code NOTBORING for a free month on Framer Pro.

Just Publish it With Framer

Bad Analogies

I saw an interaction on twitter the other day that I’m not going to share here because these specific people don’t deserve to be singled out because almost everyone is guilty of something similar.

Basically, one person said, “It’s amazing that OpenAI was able to raise $122 billion when they don’t have a single business that works,” and the other person replied, “Yeah they do they have a bunch of different business lines doing billions of dollars in revenue,” and the OP responded, “Yes but none of them is profitable, they’re losing a lot of money,” and the replier replied, “You could have said the same thing about Amazon!”

Amazon’s success has done a great deal of harm to a great number of companies.

Obviously, it’s done more good. AWS is a miracle. But go with me.

Jeff Bezos is a generational entrepreneur who came from a hedge fund and made a very calculated decision to lose money in the short term if it meant making more of it in the long term. In his 1997 Letter to Shareholders, his first as a public company CEO, he makes the trade-off he’s willing to make clear: “When forced to choose between optimizing the appearance of our GAAP accounting and maximizing the present value of future cash flows, we’ll take the cash flows.”

He spends his entire 2004 Letter to Shareholders walking investors through the logic.

Earnings don’t directly translate into cash flows, and shares are worth only the present value of their future cash flows. Working capital and capital expenditures matter too, and a company can actually destroy shareholder value by growing earnings if the capital investments required exceed the present value of the cash flow derived from them.

Bezos had his hands firmly on the wheel, and a negative working capital engine at his back. There are a million little details that made Bezos’ plan work, all of which, and the connections between them, were taken into consideration. Get to scale category-by-category, starting with the ones best suited to the internet, build an unmatched distribution and logistics network, drive down prices and delivery times, get more scale, grow the distribution network, drive down prices, make the “divinely discontent” customers discontent with anything slower or more expensive than what they’d come to expect with Amazon, get them hooked on Prime, paying a subscription and ordering more, keep pressing the advantage, grow FCF per share. All the while, thanks to negative working capital, growth generated cash, cash funded more infrastructure, infrastructure enabled lower prices and faster delivery, which drove more growth.

We all know the Amazon story to some extent, right? I don’t need to go into more detail than that for now. We could tell a similar story about Facebook: they had no revenue! they didn’t even have a business model!

You could, and many have, use this to justify all manner of sins in your business, in a business in which you’ve invested, or, in the case of the AI labs, in a business that a bunch of weirdos on the internet seem to feel is their duty to defend.

I am perhaps a bit sensitive to this after spending six years pre-not boring in the same industry as WeWork. WeWork lost a lot of money, and so people compared WeWork to Amazon, and maybe there’s an argument to be made that of course you need to spend a lot of money in real estate because you need to do construction and furnish spaces and buy a lot of beer and tequila upfront but then you have this whole network of long-term leases that will generate cashflow for years to come. Bezos wasn’t afraid of spending money today for cashflow tomorrow, why should Neumann be? But we often put in offers on the same spaces as WeWork, and we had these super finely tuned underwriting models, and looking at our underwriting models versus the prices they paid to outbid us on certain spaces it was clear that no matter how optimistic your monthly revenue projections, there was just no way they were going to make money on each space. That was before you took into account the custom builds they were fronting for customers who signed two-year deals (builds that would potentially need to be redone for the next one) or the 100% commissions they were paying brokers for an entire year’s worth of revenue on any customer they brought in. Adam Neumann went on Rick Rubin recently, and it was incredible and made me like him more and even made me bullish on Flow, but even Adam Neumann admitted that he let the economics get away from him in the halcyon days of 2018 and 2019.

If you had justified a WeWork investment to yourself by analogizing to Amazon and repeating “It takes money to make money,” you would have lost it all in the 2023 bankruptcy. Maybe the cleaned up company makes a lot more sense today than the bloated one did then, I haven’t looked at it so I don’t know, but if it does, it’s certainly not because it’s like Amazon.

Of course, analogies can be a useful starting point. “This thing that I might have otherwise dismissed out of hand is probably worth a closer look, because things that have looked this obviously wrong in the past have turned out to look pretty good” is a good way to think.

Uber, too, lost a shitload of money. Uber lost way more money than WeWork. WeWork lost like $2-3 billion in its worst year. Uber burned $9.1 billion in 2022. Uber’s losses made WeWork’s look like rookie numbers.

And yet… Uber is still a publicly traded company, with a market cap hovering around $150 billion.

If you’d looked at Uber and said, “Hey, Amazon lost money too, let me take a closer look,” and then realized that Uber was burning money in actually kind of the same way that Amazon was, to get its product as close to customers as possible in order to improve the customer experience, particularly if you’d looked during the depths of 2022 when it was burning so much cash, you could have made a smart investment.

Then again, using Uber as an analogy – literally, as in Uber-for-X – has kept the startup graveyard operators in business. The one successful company that could be called an Uber-for-X is DoorDash, and it worked explicitly because the team looked at Uber, said “there’s something interesting here,” and then tuned it to fit the situation at hand. Instead of competing in dense urban markets, they took the model to underserved suburbs, where competition was lower, CACs were lower, average order values were higher, retention was higher (fewer alternatives), pickups and dropoffs were easier (stores and homes have driveways and parking lots), and then, per Dan Hockenmaier, they executed relentlessly: “They made deliveries just a little faster and more reliable every week. They scrutinized the quality of every one of their restaurants and dashers. They optimized cost out of the system and gave it back to customers in the form of Dashpass (which launched in 2018 at the start of this chart).”

So are frontier AI labs good or bad businesses? I don’t know. Smarter minds than mine are doing a lot of non-analogical work trying to answer this question. The economy depends on it.

I do know that they’re not definitely bad businesses because they’re losing money.

They are certainly generating a lot of revenue very quickly, and growing at rates that are unprecedented at their scale. Anthropic grew from a $9 billion run rate at the end of 2025 to a $19 billion run rate in February (and rumors are it went much higher in March). In announcing its $122 billion fundraise, OpenAI said, “Within a year of launching ChatGPT, we reached $1B in revenue. By the end of 2024 we were generating $1B per quarter. We are now generating $2B in revenue per month. At this stage, we are growing revenue four times faster than the companies who defined the Internet and mobile eras, including Alphabet and Meta.”

To dismiss those numbers because of a little cash burn would be just as dumb as saying that cash burn is the same as Amazon’s.

On the other hand, it’s worth looking more closely at the burn to understand what you get for it.

Amazon was a high fixed-cost business with variable revenue. The more customers it served, the cheaper each marginal customer got. Each time you ask Claude to build you an app that tells you when to do your laundry, it costs Anthropic money. Each time you ask(ed) Sora to make you a video of Sam Altman twerking, it cost OpenAI money (and scarce compute that could be used for other things), which is why they shut it down. That said, margins at both Anthropic and OpenAI are improving, and with every point in margin, that variable cost problem gets less painful.

One thing to consider is that API tokens are likely a much better business because Anthropic charges based on tokens consumed. Funnily, when I asked Claude about this, it linked me to a public artifact of research someone else had done with Claude, which said that average users of Pro and Max make Anthropic money, Heavy users of Pro and Max lose Anthropic money, and the API business (which is 70-75% of revenue) produces “50–65% gross margins on Sonnet workloads and 35–50% on Opus workloads.” Caveat emptor. This is a monster business, but I wonder how much it benefits from another bad analogy: linesofcodemaxxxxxing.

In the olden days, lines of code (LOC) were expensive to produce. Only a select few even knew how. There were more problems to solve than people or LOCs to solve them, so companies would create a backlog and their engineers would crank through them by producing more LOCs. The lines of code themselves were not the valuable thing; they were valuable insofar as they were used to solve problems the company needed to solve. You could look at a successful software company, see that they’d written a lot of code, and lazily analogize: LOCs are correlated with success, lines of code are themselves a goal. Goodhart’s Law strikes again.

“You gotta be tokenmaxxxxxing,” they’ll tell you. I do wonder if this part of the business slows down once the psychosis breaks and we have maxxxxed all that we can maxxxx. At some point, when you’ve gone too deep, the ghost of Jeff Bezos appears and asks you what any of these tokens you’re maxxxxing has done for your customers.

Anyway, another thing about Amazon’s business is that while they had a ton of direct competitors - Barnes & Noble, eBay, Walmart, a graveyard of dot com competitors - they had no real competition in their strategy. If Bezos was right that getting to scale category-by-category, starting with the ones best suited to the internet, building an unmatched distribution and logistics network, driving down prices and delivery times, getting more scale, growing the distribution network, driving down prices, making the “divinely discontent” customers discontent with anything slower or more expensive than what they’d come to expect with Amazon, getting them hooked on Prime, paying a subscription and ordering more, continuing to press the advantage, and growing FCF per share proved to be the winning strategy, Amazon would emerge on the other side of all that spend as the only company with all of that in place, and with the pricing power and control over the levers of profitability that come with domination.

Uber had direct competition, but it was weaker, and Uber was making a very explicit bet that scale and network effects were all that would matter. They were right, too, and emerged on the other side of all that spend as the only company with the right network in place, and with the pricing power and control over the levers of profitability that come with domination.

Could you say the same for any of the AI labs?

We don’t have to spend a lot of time answering this. There is a lot of direct competition. The strategies seem to be very similar from the outside. Sure, different companies play with targeting different end users, and maybe OpenAI will turn ads into a big business while Anthropic owns the enterprise (although it looks like OpenAI is pushing all of its resources towards competing directly for coding and enterprise), but the core strategy is spend a lot on CapEx now to build increasingly strong models and maybe God is on the other side of all of that spend.

Ask any of the polytheistic religions whether competition among the gods is any less intense. And if we’re playing gods, Google is probably Zeus. Meta is Hermes - god of commerce, god of thieves, trickster god making everyone else’s shit more expensive. Should we do DeepSeek? Sun Wukong, the Monkey God who stole and devoured the peaches of immortality until the Jade Emperor begged Buddha to intervene, and Buddha trapped him under a mountain.

Our competition is multidenominational, and everyone is fighting with basically the same weapons. Every major lab, it seems, prays at the same altar, the Altar of Scaling Laws.

But! You say. But! Recursive self-improvement! Fast takeoff! Those tokens aren’t just producing code — they’re creating a proprietary data flywheel! There will be One True God.

Maybe!

Another reason I am allergic to some of this analogizing and extrapolating is because I have done it myself. I, too, have sinned. I have analogized crypto protocols to companies that didn’t fit just right. I have written I, Exponential and Compounding Crazy. As much as you think you love a good exponential, I can assure you I love them more. So when I ever dare question whether AI might hit a ceiling before it makes us irrelevant meatbags, and am met with the inevitable, “You don’t get exponentials! This is exactly like X! Have you even seen the Tim Urban chart?!” just know that I know these tricks. I have learned their painful lessons.

Not all exponentials are created equal. Does God think forever or does he know all in an instant?

I have been pushing back against Leopold’s assertion that believing ASI will come soon after 2027 “doesn’t require believing in sci-fi; it just requires believing in straight lines on a graph” since he first tweeted it. (to be fair, he’s made billions more dollars than me in the interim.)

Just this week, METR’s Joel Becker wrote a post attempting to provide an intuition for the “Straight lines on graphs” argument. The point is that even if it doesn’t make sense, the line will continue to go up. We’ll figure out ways. But my point is that even if the models can think longer, it doesn’t mean that we will birth God. I think we’re going to end up with really excellent software, which is great, instead of an all-knowing and all-loving AI.

AND, more relevant to the Amazon discussion, if each lab’s model can think longer and get smarter at roughly the same rate, without any clear runaway winner, then the situation is not analogous to Amazon, or even Uber, who ate short-term pain for long-term strategic solitude.

I don’t mean to pick on the AI labs, I don’t even think they themselves are the ones perpetrating this malanalogous thinking, it’s just that they’re the only thing anyone is talking about so they are the richest source of bad analogies. And I really don’t know how it’s going to play out. There is a chance we wake up in a decade and OpenAI and Anthropic are the most valuable companies in the world. There is also a chance that they are the AOLs of this whole thing. There’s a chance that the biggest business in a decade is the one that bought all the cheap tokens after the crash, the way Rockefeller bought all the excess rail capacity worth buying.

AOL and Rockefeller are analogies, too, huh? Avoiding them is hard, because they truly are useful starting points. They’re so easy to reach for, in fact, that on their own, they are at best consensus and at worst dangerous.

It’s the work you do from those starting points that will decide whether you end up the next Hummer Winblad or the next J.P. Morgan. (damnit)

That’s all for today. If you liked what you read, join us in not boring world:

Subscribe now

We’ll be back in your inbox tomorrow with a Weekly Dose.

Thanks for reading,

Packy

Hey friends 👋,

Happy Friday! The sun is shining here in New York City, Duke plays in the Sweet 16 tonight, and there are so many incredible stories this week that I kept having to change what made the cut as new news came out.

Grab a cup of coffee and catch up on all of it in one place. And if you have a second cup, take an hour this weekend to read Electromagnetism Secretly Runs the World.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Quince

Last week, it was cold and rainy in New York. Today, it’s warm and sunny. Aside from the renewed joie de vivre, that means it’s time to switch out the wardrobe.

More and more, when people are thinking about refreshing (and upgrading) their … anything, they ask, “Does Quince make this?” More and more, the answer is yes. Quince is the default starting point for quality, across cashmere, luggage, furniture, fragrance, fine jewelry, and new spring wardrobes.

I have been lucky to be a Quince investor for years. They recently raised at a $10.1 billion valuation. The reason is that they’re able to deliver high-quality across a growing number of categories at surprisingly affordable prices. 100% Grade A Mongolian cashmere. 100% mulberry silk. 100% European linen. 0% brand tax.

They can pull it off because Quince built an M2C (Manufacturer-to-Consumer) operating system with AI demand forecasting, real-time production planning, and direct factory partnerships that cut out distributors, wholesalers, and the retail markup entirely. This week, they even hit #1 in the App Store for shopping. The machine is humming.

Look, I’m biased, but I’m also writing this while wearing Quince jeans and a Quince q-zip. We dress our kids in Quince. And I’m about to do a big spring haul. If you haven’t started shopping at Quince, or haven’t checked out everything they have to offer in a while, join me.

Shop Quince

(1) NASA Goes Nuclear (and to the Moon)

Jared Isaacman’s NASA is shaping up to be as cool as we expected.

On Monday, Isaacman went onstage at the Hill & Valley Forum in Washington for a conversation with Founders Fund / Varda’s Delian Asparouhov and laid out his plans to build a permanent base on the Moon and send a nuclear-powered spacecraft to Mars.

In a normal week, either one would get its own top billing in the Dose, but since JI dropped both at once, let’s Start with the Moon Base. NASA is committing $20 billion over seven years to construct humanity's first permanent outpost near the lunar south pole. “The moon base will not appear overnight,” Isaacman said. “We will invest approximately $20 billion over the next seven years and build it through dozens of missions.” The plan has three phases, with Phase 1 starting now: up to 24 launches and 20+ landings in the next 35 months. That's roughly a mission every six weeks, which sounds like SpaceX cadence.

The first wave will MoonFall surveillance drones to map terrain around the south pole, fission reactors and RTGs for power, and enhanced rovers for scouting. Phase 2 adds pressurized rovers built in partnership with Japan's aerospace agency and semi-habitable modules after Artemis IV. Phase 3 is the full base: permanent infrastructure for sustained human presence. We’re going, and we’re staying. Moon will be a State.

Between NASA and SpaceX, the Moon has been getting a lot of love recently, and Mars has been pushed out. But Isaacman also announced some Mars plans.

SR-1 Freedom will be the first spacecraft to use a nuclear fission reactor for propulsion beyond Earth orbit. A HALEU-fueled reactor delivers 20+ kilowatts of electrical power, driving xenon ion thrusters via a closed Brayton cycle. It launches in December 2028 (likely on a Falcon Heavy) and should arrive at Mars about a year later. Once there, it will deploy three Ingenuity-class helicopters carrying cameras, ground-penetrating radar, and radios to scout human landing sites and hunt for subsurface water. To summarize, we’re sending a nuclear-powered space craft to Mars to release scout helicopters.

SR-1 repurposes the already-built Power and Propulsion Element from the scrapped Gateway program. Rather than shelve expensive hardware when priorities shifted, Isaacman's team pointed it at Mars. The reactor will activate within 48 hours of launch, then it’s off to the Red Planet.

Nuclear is going to be critical to our space ambitions, whether going or staying. We wrote about it in our Deep Dive on Radiant. Fission reactors will power the Moon Base and a fission reactor will propel SR-1 to Mars. Eventually, we’ll have nuclear-powered civilizations as far as the eye can see.

Looks like putting Isaacman in charge of NASA set off a … nuclear chain reaction.

(2) Terraform Industries Breaks Ground on Synth Hydrocarbon Site

Casey Handmer on TBPN

We’re going to need a lot of energy here on earth, too, and even as we add a ton of clean nuclear, solar, and even geothermal, we’re going to need those sweet, sweet hydrocarbons. We just want them to be clean, too.

Good news. Casey Handmer and the Terraform Industries team have been hard at work figuring out how to turn sunlight and air into pipeline-grade natural gas for less than the cost of drilling it. In March 2024, his team in Burbank proved the chemistry works by producing synthetic methane end-to-end for the first time. Now, two years later, Terraform Industries has broken ground on a manufacturing site in Kern County, California, to scale up production.

Kern County is a lucky place: it’s one of California's biggest oil-producing regions and it's also absolutely sun drenched. Those rays are what Terraform wants. The Terraformer system is straightforward in concept: solar electricity powers water electrolysis for hydrogen, direct air capture pulls CO2 from the atmosphere, and a reactor combines them into carbon-neutral methane. What's hard is making it cheap enough to compete with drilling a hole in the ground. Terraform's bet is that plummeting solar costs will get them there, and their qualified electrolyzer stack (under $100/kW) and full-scale reactor suggest they’re heading in the right direction.

Last week, we wrote about one Australian, Chris Power, launching a huge new manufacturing site in Alabama. This week, another Aussie, Handmer, is launching a hydrocarbon manufacturing site in the California desert. Chris and Casey are two of the smartest people I’ve met. With them working on the country’s behalf, it’s a very g’day in America.

(3) Arbor Energy Lands Billion-Dollar Turbine Deal

TechCrunch

But wait… how are we going to turn all of our natural gas into power to feed the data centers? I’m glad you asked.

Arbor Energy, founded by former SpaceX engineers Brad Hartwig and Andres Garcia Clark, just landed a deal with GridMarket to supply up to five gigawatts of its Halcyon turbines, or roughly 200 units, valued in the single-digit billions. They expect the first turbine to connect to the grid in 2028 to be manufacturing 100+ per year by 2030.

Turbines are so hot right now. They’re such a big bottleneck to the data center buildout that their shortage is a big driver behind the rush towards orbital data centers, where they can be powered by the sun. Traditional gas turbines from the big OEMs are backordered until 2032. Hartwig puts it simply: “Everyone wants more power. They wanted it yesterday. The time frames are compressing and the scale is getting larger.”

Arbor takes rocket turbomachinery designed for spaceflight, 3D-prints it into 25-megawatt modular turbines, and sells them to the data centers that needed power yesterday. The turbines run a supercritical CO2 cycle with oxy-combustion, which means zero operating emissions on natural gas, or carbon-negative operation if you feed them biomass waste. They're fuel-agnostic, pre-assembled, and 3D-printed, which is important because it decouples manufacturing from the supply chain bottlenecks (specialized blades, vanes, castings) that plague conventional turbine production or even alternatives proposed by companies like Boom Supersonic.

Our friends at Cantos can’t stop winning.

(4) Centivax raises $37M to advance universal flu vaccine

Katherine Davis for Axios

All of this progress in rockets and energy is sick… but you might not have to be.

Last month, we covered research on a single vaccine to protect against all colds and flus. The company behind that research, Centivax, whose homepage reads “smash the mutants,” just announced that they’ve raised $37 million in follow-on funding led by Structure Fund, with Meiji Seika Pharma, Sigmas Group, Kendall Capital Partners, and Stripe founders Patrick and John Collison joining. The company has now raised $133 million, including from our friend at Amplify Bio.

When we get the flu vaccine every year, we line up for a shot that's been reformulated to match whichever strains scientists think will circulate that season. Sometimes they guess right and sometimes they don’t. Centivax wants to change that with a single dose that protects against all flu strains, with a booster needed only every two to four years instead of annually. The company's universal flu vaccine is in a Phase 1 trial, with results expected by end of year.

“If our data looks good by the end of this year,” CEO Jake Glanville said, “effectively, the pandemic era for influenza is over.”

Centivax is applying its approach, targeting conserved regions across strains rather than trying to keep up with seasonal mutations, beyond flu, too: universal snake venom, Alzheimer's, malaria, and cancer vaccines are in the pipeline. As Glanville puts it, “You want to treat all the flus, not just one strain. You want to treat all the snakes, not just one snake.”

Of course, it wouldn’t be a medical miracle entry without some FDA hair. Last month, the FDA about-faced on Moderna’s mRNA flu vaccine proposal and put it back under review. Just to be safe, Centivax is hedging by running trials in Europe and Australia alongside the US. “If something unexpected happens with the FDA, we would just proceed in Europe,” Glanville says.

The possibility that Europeans get a universal flu vaccine before Americans because of regulatory dysfunction is both absurd and completely believable. Hopefully, if the FDA blocks the universal everything vaccine, they will also watch my kids during sick season.

(5) How to Turn a Chicken Egg Into a Drug Factory

Carl Zimmer for The New York Times

I am so clucking eggcited that Neion Bio is finally coming out of its shell.

Today, pharma uses Chinese hamster ovary (CHO) cells to produce biologics like Keytruda and Humira in huge stainless steel bioreactors. Merck recently spent $1B on a single Keytruda facility.

In early 2024, Elliot wrote that chicken eggs are much more efficient bioreactors. They run on grain and water, produce six grams of protein per unit, and we already farm them at massive scale.

As the (excellent) Neion website says, chicken egg drug manufacturing can be “extremely low COGS, hyper resilient, CapEx avoidant, ultra scalable, very high reliability, and lower environmental impact.” Nature evolved the chicken egg to be that way. The hard part has been that the science hasn’t been able to modify what’s inside the egg fast and reliably enough to actually manufacture drugs.

Sam Levin, who we previously backed at Melonfrost, and Dimi Kellari explored the frontiers of chicken egg research, which they could get their arms around because the cutting edge stuff is happening in a very small number of labs around the world. They went to all of them, and they realized that the time was right to build a company that uses nature's bioreactors to produce drugs at a fraction of the cost.

In today's NYT article, Sam predicts that the cost can be 1/10th or even 1/100th of the current cost, and that just 3,900 hens could meet global Humira demand.

I'm proud to back Sam, Dimi, and the Neion Bio team out of not boring capital as they work to hatch the balk of the world's biologics and dramatically lower the cost to produce critical drugs, right here in NYC. And I'm sure they're happy that I can share my chicken puns with all of you instead of just replying to investor updates with them.

Extra Dose: Science Breakthroughs, GPS, Memory Without Brains, Astro Mechanica

Join not boring world to get an even bigger Dose of Optimism.

Read more

Subscribe now

Hi friends 👋 ,

Happy Tuesday! Welcome to our newest installment in what has become an unintentional two-part series on non-LLM models that can do things that humans can’t, things that will give us superhuman abilities in the physical world. They’re also both co-written with founders you’d expect to find in SF but are building right here in the greatest city in the world, NYC.

The first was last week’s essay on World Models with Pim de Witte.

Today’s is about machines that can intuit electromagnetic fields in a way almost no humans can that will help us design and build better electromagnetic (EM) systems.

As you know, I’m very bullish on the growing role of electromagnetic systems in the economy. After Sam and I wrote The Electric Slide, Arena Physica CEO Pratap Ranade and I traded emails. In one of them, he wrote:

The electrical and electromagnetic components are the “nervous system” of modern hardware and contribute to 40-50% of failures. Our ability as a nation to test and build it has declined, but –– imo even bigger –– as a species, we’re still unable to wield electromagnetism to its full potential.

Over the past seven months, we’ve developed a friendship, and Pratap has broken my brain many times. One of the things that’s most fascinated me is the idea he’s betting his company on, the one he emailed me about: humans can’t intuit EM, and it’s a bottleneck to the electric progress we both want to see. There’s no reason machines can’t be taught to understand them much better than we can, though.

For the past few years, Arena has been building AI tools and deploying expert electrical and RF engineers to help companies design, develop, and debug electromagnetic hardware. They’re working with companies including AMD, Anduril, and Sivers Semiconductors. They are backed by investors including Founders Fund, Peter Thiel, Initialized (Garry Tan), Shield Capital, and 137 Ventures.

Today, they’re re-branding as Arena Physica with the expanded mission to develop “Electromagnetic Superintelligence.”

This is an essay about how to teach machines to see the fields that we can’t, and what the world might look like if we can.

Let’s get to it.

Today’s Not Boring is brought to you by… The Pitch by Deel

Normally, our sponsors would like me to tell you why you should give them your hard-earned money. This time, Deel has asked me to tell you how they can give YOU money in The Pitch.

Deel recently launched a global tournament with $15M in prizes for startups: the top 10 get a $1M investment each, and 100 regional winners will get $50k. You don’t need warm intros and you don’t need to pay to apply. It’s just a pure competition for the best entrepreneurs.

Pitch in just two minutes for the chance to win a $1M investment, access to a global ecosystem designed to help you scale (partners and sponsors include Stripe, Google, AWS, and a16z), and to get your startup in front of global leaders. This is your shot.

Apply Now

Electromagnetism Secretly Runs the World

Electromagnetism secretly runs our world. “Secretly,” because only a few people on this planet can intuit how it works.

Your phone’s GPS is powered by satellites that broadcast electromagnetic (EM) waves with timestamps. The wi-fi in your apartment is created by EM waves bouncing around the walls. Air traffic control is radar, as in EM waves that pulse out and listen for echoes off aircraft. When Maverick locks onto a bogey in Top Gun, he’s using a phased array radar steering EM beams electronically. Contactless payment? EM. Microwave oven? EM. The fiber optic cables carrying the internet across the ocean floor and through Somos’ network? That’s light… which is also EM.

Every single wireless signal, medical image, radar sweep, every chip talking to another chip inside a data center. All of it is electromagnetic waves, shaped and directed by physical structures designed to manipulate these waves. And, as electricity and intelligence race to define our era, EM’s presence is only growing more pronounced. In our data centers, chips communicate with each other via short-range EM waves. If Elon successfully moves the data centers to space, AI will be beamed down from satellites to your device via EM waves.

As Packy and Sam wrote in The Electric Slide, everything that can economically go electric will. Cars, trucks, buses, drones, boats, stoves, heat pumps, batteries, bikes, even planes, anything that moves, heats, lights, computes, or converts energy is moving from mechanical to electric. All of those newly electric things will be full of EM components.

In 1970, electronics accounted for five percent of a new car’s cost, on average. By 2020, that number reached forty percent. By 2030, it’s anticipated that the cost of the electronics of a consumer automotive vehicle will reach fifty percent of the vehicle cost.

Electronics comprise 35% of the cost of the F35 Lightning II, more than the cost of the engine itself, and 15% of the Pratt & Whitney F135 engine, which costs $20 million. By the 2030s, when it’s projected that defense contractors will be building the F-47, they’ll be spending over 40% of the $300 million airframe on electronics.

This is good. We want to see this electrification continue. Electric machines perform better with less impact on the environment, give us capabilities combustion engines can’t, are better-suited for autonomy, and are riding cost/performance curves that will continue to widen the advantage.

But among a number of challenges addressed in The Electric Slide particularly as it relates to production, there’s an equally large one looming in the research and development of new and better electromagnetic machines: our electromagnetic capabilities are bottlenecked by the very small number of humans who actually understand how any of this works.

There is a reason radiofrequency (RF) engineering — the practice of designing hardware that shapes and directs EM waves — is often referred to as black magic. There are maybe ten people in the world who can deeply intuit electromagnetism, who can see which shapes will create which EM fields in their mind’s eye1. I am not one of those people, but I’ve met them. I’m hiring them at my company, Arena Physica, and I went to school with many of them.

There was a guy in my physics program about whom our professor asked, “You know what’s special about this guy?” We all said no. “This guy thinks like an electron.”

What he meant was that, electrons, if they were sentient, would feel all of these different fields pulling at them. Electrons would probably have an intuition for this feeling, the same way we have an intuitive feel for gravity, how we just know that when we let go of a ball, it will drop to the ground. Those friends of our ancestors’ who couldn’t intuit gravity did not live long enough to reproduce.

Some people – a vanishingly small number – have spent enough time studying, testing, designing, and simulating electromagnetic systems to be able to intuit them like gravity. But for the rest of us, electromagnetism is mostly invisible.

For the vast majority of human history, we haven’t needed to see beyond the visible spectrum in order to survive. And so we haven’t. Those of our ancestors’ friends who wasted precious resources on seeing the full spectrum of EM waves wouldn’t have lived to pass on these traits, either.

Humans can see a small part of the electromagnetic spectrum, the “visible light” portion with wavelengths between 400 nanometers (violet) and 700 nanometers (red). We don’t see shorter wavelengths (ultraviolet light, X-rays, or gamma rays) or longer wavelengths (infrared, microwave, or radio waves).

This has served us just fine. Until electromagnetism came to run the world.

We have a fundamental force that we rely deeply on, but one that very few of us can work with naturally. This slows technological progress and limits what we can make.

Fortunately, AI doesn’t share our blind spots. It is particularly good at seeing patterns, at making connections and understanding dependencies that are not necessarily intuitive to humans.

Because of this, we believe that computers will be much better at grasping electromagnetism than we are. We should be able to build a Large Field Model (LFM) — like an LLM that generalizes across language, except ours generalizes across EM. We should be able to use this LFM to understand EM waves and shape them to do what we’d like them to do.

That’s the big bet we’re making at Arena Physica. To understand why we’re making it, I want to first make sure you understand electromagnetism.

A Brief Primer on Electromagnetism

Packy and Sam gave A Brief History of Electromagnetism in The Electric Slide.

I’m going to add to that with A Brief Primer on Electromagnetism. I’ll sprinkle in relevant history, but my goal is to make sure we have a working understanding of electromagnetism.

There are four fundamental forces that govern how everything in our universe works:

1. Strong force

2. Weak force

3. Gravity

4. Electromagnetism

The strong force and weak force operate at subatomic scales. The strong force binds protons and neutrons together in atomic nuclei. The weak force enables radioactive decay and nuclear fusion.

Gravity is the weakest of the four forces by an enormous margin (roughly 10³⁶ times weaker than electromagnetism). Yet, it dominates at cosmic scales. It only attracts, never repels, which means its forces keep adding up. And it acts on every single particle with mass or energy. It’s also mysterious at a fundamental level: gravity and quantum theory are incredibly powerful theories for how our world works, but they are fundamentally incompatible. This remains one of the deepest unsolved problems in physics. But we have a strong, intuitive relationship with gravity. In our day-to-day lives, we can feel the force.

Electromagnetism is the force we interact with most directly in everyday life. It’s also the one we’ve industrialized most aggressively. It governs light, electricity, magnetism, and chemistry, essentially everything about how matter behaves above the nuclear scale. It’s why matter has structure, why chemistry works, and why technology works. It’s responsible for the structure of atoms (electrons bound to nuclei), the bonds between molecules, the rigidity of solid objects, and all electronics and communication technology. Unlike gravity, electromagnetism has both positive and negative charges, which means it can attract or repel, and large accumulations tend to neutralize themselves. Our mathematical understanding of electromagnetism is extremely accurate: it is described by quantum electrodynamics (QED), the most precisely tested theory in all of science.

And yet… despite this precision, electromagnetic systems can be deeply counterintuitive. RF engineering, for instance, has a reputation as black magic. The wave-like, distributed nature of fields at certain frequencies produces effects that violate the intuitions built from simple circuit theory.

But let’s try to build up our intuition as best we can.

Every force carries energy. Electromagnetic energy comes in quanta we call photons, particles of light, but for most of what we build — antennas, radars, communication systems, phased arrays — it is easier to think about it as waves, in terms of frequency, wavelength, and phase. A photon has different amounts of energy based on its frequency, which can be seen on the electromagnetic spectrum. A super-high energy photon would be something like an ASML machine making chips using EUV (extreme ultraviolet). EUV is very, very high frequency. Therefore, it is very high energy, and in turn, very short wavelength. As you go through the visible spectrum and get to the other side, you reach infrared, and electromagnetic energy becomes heat. Then you get to RF (radiofrequency). With RF you have very low-energy photons. But it’s really just all electromagnetic waves that follow the same principle. High energy = high frequency = short wavelength and vice versa. (Explore the EM spectrum here.)

Now, think about a prism. A prism is an object or a material that treats incident EM waves — or different photons — differently. So if you’re a red photon, your refractive index, or how much you bend, is a certain number. If you’re a blue photon, it’s different. Refract the whole beam and you end up with that beautiful rainbow on the other side of the prism.

The prism is an early example of humans manipulating electromagnetic fields. People noticed that light, when it passed through glass or a crystal shaped in a certain way, would form a rainbow. We’ve come up with many ways to manipulate electromagnetic fields since, the most consequential of which is also the simplest.

If a caveman discovered he could manipulate electrons, the very first thing he might do is the simplest possible thing: electron on, electron off. On, off. One, zero.

The prism is actually much more sophisticated than the on, off switch. It’s more subtle, more expressive, and its implications are more powerful. But computing is based on the caveman’s idea of flipping the switch.

Early computers were literally built on this simple idea: they used mechanical switches known as relays to compute. When the metal touches, current flows (1/on). When it separates, current stops (0/off).

One problem with mechanical switches was that, when you switch often in air, the air ionizes and creates tiny bolts of lightning that can jump across the gap and create an arc. These arcs can make your “bit” unreliable. Sometimes it’s supposed to switch, and it doesn’t.

The other problem was that they are very slow. It was the advent of the electronic switch that put us on a path to be able to now build processors that can work at Ghz speeds (10⁹ cycles per second).

So we invented vacuum tubes2. They removed the air. Without atmosphere, there was no arcing. But vacuum tubes were fragile, power-hungry, and couldn’t scale. The great breakthrough was the semiconductor, materials like silicon that can have their conduction controlled. They can be made to conduct or not conduct (hence semi-conductor) based on applied voltage. Semiconductors enabled the transition from mechanical to digital, and it gave us the transistor, which are tiny silicon devices that switch on or off using voltage. This gave us Moore’s Law, which gave us Boolean logic, which gave us everything in modern computing. That singular innovation, the transistor, has produced most of our technological advancement over the past seventy years.

But if you read Gordon Moore’s 1965 paper, the one in which he described what would come to be known as Moore’s Law, you’ll find that only the first half is about digital silicon; the second half is about analog silicon.

Nobody paid attention to the analog part, but I think it’s even more fascinating today than the digital one.

Digital silicon is about switching: transistor on or off, conducting or not, one or zero. All the gates, all the logic, all of computation follows from that binary foundation. It’s powerful, but it’s also, as we’ve discussed, the simplest possible thing you can do with an electron. It’s caveman math.

Analog silicon is about shaping. Instead of just on/off, you’re asking: what if I could bend the electromagnetic wave? What if I could guide it, direct it, absorb it at specific frequencies and reflect it at others? In practice, this is RF front-ends, antennas, packages, and printed circuit boards (PCBs) behaving like a single, unified electromagnetic object.

This is how the world works too. The world is analog. The world does not work in 0s and 1s, but rather in the continuum in between them. Even if all computation is done digitally, you’ll need to deal with analog signals and shape waves the moment you need to interact with the real world (for example, capture sound in a microphone, produce sound in a speaker, send wireless signals over the air, send light on optical fibre)

Remember the prism? That’s what analog silicon does, but for all electromagnetic frequencies, not just visible light. Instead of glass bending light, you can use carefully shaped conductors printed on silicon to bend, direct, and shape EM waves.

This is where we leave the realm of deterministic computing and enter a world of black magic.

Try This At Home

Here’s an experiment. You can try this at home.

Materials List: COPPER WIRE, COMPASS, BATTERY.

Take your copper wire, connect it to a battery, and run current straight through it. The magnetic field it produces will wrap around the wire in a helix. You can confirm it’s doing this by holding a compass near it and watching the needle deflect perpendicular to the wire.

Now, coil that same wire into a spring shape (a solenoid) by wrapping it around a pencil or screw 10-15 times. Run current through it. The magnetic field is completely different: instead of wrapping around the wire, it shoots straight through the center of the coil. Same wire, same current. But a different shape = radically different field.

This is the fundamental game of electromagnetism: geometry determines behavior. Every antenna, radar, or phased array tile is just a more sophisticated version of this principle. Find the right shape, and you can make electromagnetic fields do almost anything.

To understand why shapes matter so much, consider what happens when an electromagnetic wave hits a conductor.

A conductor is special because it has free electrons. Free electrons are not locked into a lattice like in an insulator, but instead swim around in what physicists call an “electron sea.” When a photon (an EM wave) hits this electron sea, those electrons start to move in response. They oscillate with the wave.

This is fundamentally how an antenna works. The old bent TV antenna on your grandparent’s roof was shaped specifically to receive UHF frequencies broadcast from a distant TV station. The EM waves traveling through the atmosphere would hit the antenna, excite the electrons in the metal, and those oscillating electrons would travel down the wire into your TV as a signal.

That signal carried information, like encoded images of I Love Lucy, compressed into patterns of electromagnetic oscillation, broadcasted through the air, absorbed by your antenna, decoded by your TV. If you step back and think about it, this entire chain is completely absurd. We transmit moving pictures through the air using invisible waves. And turning those waves back into pictures all comes down to the shape of a wire.

Radar works basically the same way, except it’s more high-powered and moves in reverse.

World War II accelerated radar. It also showed us how badly we needed it. The Allies were being pummeled and needed to track incoming threats. They turned to radar, which had developed quickly thanks to the war. In the late 19th and early 20th centuries, Heinrich Hertz (of Hz fame) showed that radio waves could reflect off objects. Several physicists also noticed that radio signals behaved strangely when ships or other objects were nearby. Through the 1920s and early 1930s, scientists in the U.S., U.K., Germany, France, the Soviet Union, Italy, and Japan all experimented with using radio echoes to detect objects.

In 1935, a Brit named Robert Watson-Watt (no relation to the steam engine Watts) proposed and then demonstrated a practical aircraft-detection system using pulsed radio waves. This led to the Chain Home early-warning network along the English coast. Chain Home was operational at the start of WWII and gave the Royal Air Force so much of an early warning in the Battle of Britain that it’s often credited as a key factor in preventing a German invasion. The United States picked up development a bit later, with the benefit of British tech transfer, and scaled up the technology’s capabilities and manufacturing. In the U.S., Alfred Loomis led research efforts at Tuxedo Park3 and helped establish MIT’s Rad Lab, which developed fire-control radar, airborne radar, and navigation radar. Germany built parallel systems that pushed the state-of-the-art in different directions.

Radar, instead of receiving a broadcast signal, transmits a beam on multiple wavelengths, waits for it to reflect off something (like a bomber), and then listens for the echo. If the object is big enough and close enough, you can detect it.

But to scan the sky, you need to point your beam in different directions. In the 1940s, that meant literally spinning a large dish antenna. You needed mechanical motors to do it. A mechanical gimbal rotating a giant antenna.

This worked, but it had obvious limitations. Moving parts break, for one. For another, the dish can only spin so fast. Today, Starlink satellites need to update their pointing multiple times a second, since they are moving at 7.6km/second. Try doing that mechanically and for 5,000 simultaneous beams.

This is where the second half of Moore’s 1965 paper becomes relevant. Moore realized you could use transistors to solve the spinning-dish problem. Replace the mechanical movement with electronic steering.

The key insight is constructive and destructive interference, the same phenomenon you see when ripples on a pond meet and either amplify or cancel each other.

Imagine you have a grid of small antenna tiles instead of a single big dish, like a checkerboard where each square is a tiny antenna. Each tile can emit a signal. Each tile is the whole RF front-end and antenna structures across chip/package/PCB that acts like a single, holistic EM object. Now, if you start the signal from the leftmost tile first, then the next tile a tiny bit later, then the next, and so on and so forth, the wave fronts from each tile will interfere with each other. Get the timing right and they’ll constructively interfere in one specific direction, creating the effect of a single, focused beam pointing in the direction you want.

Change the timing pattern, and the beam will point somewhere else. You can replace moving parts with analog and digital logic that controls when each tile fires.

This is called a phased array. And it’s how modern radar works. If you want to develop a better intuition by playing with it, we built a little simulator here.

The radar on an F-35 is called an AESA (Active Electronically Scanned Array). Nothing on it moves. It’s just a grid of semiconductor tiles, and the “beam” sweeps across the sky purely through timing. It’s also how Starlink works. Each Starlink terminal has 1,280 of these beam-forming silicon tiles. That’s why you can buy a flat panel for $300 that does what used to require a million-dollar spinning dish.

What’s happening on those tiles?

Remember: digital silicon is about transistors switching on and off. But the tiles in a phased array shape electromagnetic fields through their physical geometry.

Think back to the TV antenna, a bent wire specifically shaped to receive certain frequencies. Imagine you could print that shape onto a silicon chip, layer by layer, laying down copper traces in specific geometries, creating structures that interact with EM waves in precise ways.

On one layer, you might have a spiral. On the next, a grid. On the next, something that looks like a QR code. Stack them up with tiny connections between layers called vias, and you’ve created a three-dimensional structure that can emit, receive, absorb, and reflect electromagnetic waves at specific frequencies and in specific directions, which you have full control over.

This is what a phased array tile actually is: a 3D sculpture of copper and silicon, designed so that electrons create exactly the EM field you want when electrons flow through it.

Why This is So Hard to Build

Creating an EM field of your liking is all about geometry, which means it’s all about shapes. But how do you know which shapes to make?

With digital silicon, the rules are relatively simple. Transistors are either on or off. You can simulate billions of them with perfect accuracy. The design problem is about routing and timing, but the physics is well-behaved.

Analog silicon is different. The physics is wave physics, and waves do things that violate our intuitions.

At optical (high) frequencies, we often get away with “ray optics” intuition; light mostly travels in straight-ish lines, reflecting and refracting, and you can treat its components as fairly local.

At RF, the wavelength is big enough that your whole device becomes part of the circuit. Fields couple into enclosures, PCBs, screws, nearby tiles… Everything talks to everything. That’s why RF feels like black magic, and why you have to simulate the whole object to predict how it’ll perform.

When you’re designing a Starlink tile, for example, you can’t just model the tile in isolation. The EM waves emanating from it will interact with the entire Starlink unit: the metal casing, the other tiles, the mounting bracket, the support structure. You have to simulate the whole system at once.

This is why there are no automated tools for analog circuit design. Digital circuits can be “synthesized” from code; a digital designer can write “RTL” code that describes how his digital circuit works. Then, there are tools that can read the code, and “compile” it to a chip. But no such tools exist for analog design. There are no “standard cells” for analog, no standard analog designs, no standard building blocks. Everything interacts with everything.

Which is why there’s no ARM for analog silicon4. There are no companies that can sell “IP,” standardized circuit designs, to a multitude of customers in a highly profitable fashion — no such standard circuit exists. Every new system is different, and as a result, each new customer has different needs.

ARM can design a chip that works in any phone because digital chips are self-contained. But an analog phased array tile designed for a Starlink terminal won’t work in a different satellite. The interference patterns will be completely different!

And the simulation tools are slow. The equations governing electromagnetic fields are called Maxwell’s equations, four partial differential equations that are notoriously difficult to solve.

They’re just equations. What’s the problem?

If an EM wave is at a higher frequency, it’s more “particle-like”— intuitively, it’s like a ball — you know where it is, and it bounces off stuff cleanly. If the ball is in one corner, it doesn’t affect anything in the other corner. As EM wavelengths get longer (into RF), they become more wave-like, and the particle is sort of “spread out.” The waves start to interfere with each other a lot, like ripples on a pond. They can either strengthen or cancel each other out.

So, if you’re NVIDIA and selling high frequency chips in boxes, you can sell a single product. You can design one GPU and sell it to everyone. There is no difference between the Navy putting your chip on a ship, or Sony putting it in a Playstation. They can all just buy the chip and plug it in. But, for example, if you’re buying components for a phased array system, you have to model the entire system, because it’s type 2, not type 1. A design for a Navy ship won’t work in a Starlink terminal. The EM fields interact with everything around them—the metal casing, the mounting structure, nearby components. Change the environment, and you need a completely new design. Everything becomes a custom services problem that is rate limited by these rare experts and simulation.

In short, the solvers are slow, even with supercomputers or programs like Ansys, because the equations are tough to solve and require expertise to wield. The reason the equations are very hard is boundary conditions (the edges, where smooth calculus breaks down), e.g., a sharp metal edge can cause problems by creating strong electromagnetic reflections that cause fields to concentrate in unexpected ways.

Running a full simulation of a proposed design can take hours or days. Here’s an example design loop: make your best guess at a shape, wait hours for simulation, discover it doesn’t quite work, adjust the shape, wait hours again. As we collaborate with field experts, we witness each simulation iteration with legacy tools taking a week. That’s not enough ‘shots on goal’ to develop Electromagnetic Superintelligence.

RF design can’t be done by brute-force computation. The search space is infinite and each evaluation takes too long.

Take a simple two-layer circuit where each pixel in a 64×64 grid can be either metal or dielectric. That’s already roughly 2^64x64, or 10^1,233, possible configurations for a single, small component. The entire history of human RF design has explored a vanishingly small fraction of that space, obviously. Here, see how many of those configurations you can come up with.

Navigating this search space requires intuition. You need someone who can look at the desired field pattern and just... sense what shape might produce it.

The people who can do this have spent decades building up a feel for how electrons move through structures, how fields bend around corners, how waves interfere. They can sketch a spiral on a whiteboard and tell you roughly what frequencies it will emit strongly and which it will absorb. Aside from my classmate who could see like an electron, though, this intuition isn’t natural even to those special few. They acquire it painstakingly over long careers. Because, unlike gravity, there was never evolutionary pressure to understand electromagnetic fields outside of the visible spectrum. We don’t feel them. They’re invisible to us.

I watch my baby daughter learning about the world. She already has an intuition for mechanics. She knows that if you roll a glass off a table, it will break. She has no intuition for electromagnetism, which is probably genetic. 99.99% of people don’t.

It is a miracle that we’ve been able to manipulate EM waves to our purposes to the extent that we have. But the world is only getting more electromagnetic, and we will need a lot more shapes.

That means we need to build something that does have intuition for electromagnetism.

AlphaGo for Electromagnetism

In 2016, DeepMind’s AlphaGo defeated Lee Sedol, one of the greatest Go players in history.

The moment that stuck with everyone was in Game 2, Move 37.

The expert commentary went something like this: “That’s a mistake.” Then: “That’s stupid.” Then: “That’s a very strange move.” And finally: “That’s beautiful. That’s elegant.”

AlphaGo had done something no human would have tried, a move so unconventional that the world’s best players initially dismissed it as an error. But it worked. The machine had discovered a strategy that humans, despite thousands of years of playing Go, had never found.

What made AlphaGo possible? Two things. First, Go has clear rules and a perfect simulator. You always know exactly what state the board is in and which moves are legal. Second, because of these limitations, a computer can play millions of games against itself very quickly. AlphaGo learned by playing more games of Go than all humans in history combined.

We want to do what AlphaGo did for Go — but for physics. What if we could build a system that played millions of “games” of electromagnetic design and developed an intuition that humans simply can’t acquire?

There was an obvious obstacle between us and that dream. AlphaGo worked because Go is a perfect simulation. You know exactly what happens when you place a stone on the board. But physics is more complicated, and the simulators are slow. Maxwell’s equations take hours to solve. You can’t “play a million games” overnight.

So we needed to build the simulator first.

The EM foundation model that we’ve built, and continue to scale, is the simulator for EM physics.

The starting point for what we’ve built is known as a neural surrogate. The idea is simple: instead of solving Maxwell’s equations from scratch every time (which is slow), you train a neural network to approximate the solutions (which is fast). It’s like the difference between calculating the sine of an angle by hand versus looking it up in a table, except the “table” is a neural network that can interpolate to angles you’ve never seen before.

Traditional physics simulators work by brute force. They break space into tiny chunks, apply the equations at each point, and iterate until the solution converges. It’s accurate, but a single simulation can take hours.

But what we’re building goes beyond a surrogate. Most surrogates in physics are narrow: trained to approximate one specific simulator for one specific class of problems. Arena Physica’s model learns the relationship between shapes and fields directly, allowing it to generalize. It’s not a faster calculator (it’s not a calculator at all). The neural surrogate is learning the syntax of physics. Just as GPT learned the “logic” of language, our model is learning the “logic” of fields. Show it enough examples of “this shape produces this field pattern,” and it learns to predict new patterns for new shapes almost instantly. We’re talking about 18,000x speedups, hours to milliseconds.

If you’re reading closely, you might be thinking to yourself: of course you can go faster if you’re just trying to get an approximate answer and not a perfect solution.

Good catch. This is where the magic happens.

When you’re searching for good designs, speed and direction matter more than precision.

Think about how an experienced RF engineer actually works. They use their intuition to filter out ideas that probably won’t work and to get the rough shape of ideas that might. Then, they simulate those. They make fast, approximate judgments to decide where to invest their slow, precise simulation time.

Arena Physica’s model does the same filtering, just much faster. It doesn’t need to tell you exactly how well a shape will perform. It needs to tell you how each shape will perform relative to the others. Good enough for search is a much lower bar than good enough for publication.

Speed lets us flip the problem around. Instead of asking “what field does this shape produce?” we can ask “what shape produces this field?” That’s generative design. We specify what we want, say, an antenna that transmits strongly at 28 GHz but rejects interference at neighboring frequencies. The system uses our desired state to generate shapes that might achieve our goals.

Then, we pair two models in a loop: one that generates designs, and one that evaluates them5.

The generator proposes a batch of shapes, many of them wild, strange, things that no human would come up with. Move 37 shapes. The evaluator characterizes them all in seconds, directionally: this one’s terrible, that one’s promising, this one’s interesting. The best candidates get refined via small variations and perturbations. The evaluator evaluates the refinements. Repeat.

At each step, we ask, “Is this one better for my goal than the shape we had before?” Because we know what the rules are, like AlphaGo, and what our goals are, like AlphaGo, we can reward the model for getting closer. And like AlphaGo, by making simulation cheap, we can explore much more of the design space than we could have if everything we wanted to try required a precise multi-hour simulation.

In Asymmetry of verification and verifier’s rule, OpenAI’s Jason Wei describes the “verifier’s law.” Essentially, it says that any task that’s easy and fast to verify will be automated by AI. The hard part about our world is verification relies on special humans and simulators that are slow and expensive. By attacking this first via the foundation model for fields, acting as a fast simulator, we’ve made this problem accessible by generative AI for the first time. Our generator learns the weights through our feedback loop.

This is the same loop that made AlphaGo work: generate, evaluate, learn, repeat.

Run the loop for yourself here. You get a visceral sense for the importance of speed.

Of course, this loop only works if you have enough training data to feed it, and unlike LLMs, which can scrape the internet for training data, EM field simulations don’t exist in the wild. Nearly every single data point has to be created. So we’re building our own Data Factory.

To do that, we’re hiring the best RF lead designers we can find. Put them in one place, where, theoretically, we can amortize the capabilities of this scarce group of people over all sorts of existing and newly possible customers. Have them create designs, give feedback, test their designs, run through the loop.

We generate random designs synthetically, our experts create seed designs that our system can then procedurally amplify, and we fabricate the top candidates and pipe real-world measurements back into training.

The Data Factory has three layers: high-volume synthetic data, high-information expert-seeded data, and ground-truth fabrication data.

Because of how manual the Data Factory is today, we need to pick our use cases strategically. We are starting with analog silicon (chip packaging, phased array components, RF front-ends) and the full phased array system we mentioned above. We’ll be expanding into new domains, like superconducting quantum computing, in conversation with our partners.

The factory is the moat. You can’t build a foundation model for EM without it, nobody else that we know of is building one, and to try, you’d have to hire from the very small pool of experts that we’re bringing together at Arena Physica.

Then, when we’ve converged on something promising, by feeding the Data Factory’s output into our loop, we validate it by running the slow, precise traditional solver on our best candidate. Or better yet, by fabricating the design and validating it in the real world.

Remember why there’s no ARM for analog? At RF frequencies, the wavelengths are long enough that EM waves interact with everything around them. When designing a phased array for a Starlink terminal, you can’t just model the chip; you have to model the chip, the circuit board, the metal casing, the mounting structure, everything. It all affects how the EM waves behave.

Because of that, we’re even using our own components to build an entire phased array system for imaging and detection from scratch.

We’ll tape out silicon before the end of the year. And, for any part of the problem that’s not analog, we’re actually using our agentic stack — our hardware-aware agents, operating on our metagraph – a dynamic graph representation of the hardware – talking to tools via MCP — to speed up every single aspect of the process, so that we can go end-to-end faster. In this way, we benefit from all the amazing leaps coming from the foundation models, but also own something they can’t replicate: an EM foundation model, fed by our own Data Factory.

The system compounds: fast approximate evaluation enables broad search, broad search finds promising candidates, fabrication validates and generates training data, training data improves the generator, better generator enables even broader search.

If you can do all of the RF design work in our loop, you can build an analog IP Factory.

The IP Factory for RF and the Compiler for Atoms

Our EM Foundation model’s key advantage over existing surrogates is that it can generalize.

Take a philosophical leap with me.

LLMs don’t mechanically learn to classify a sentence. They train on the structure of words and sentences in relation to each other. The rest of their behavior is emergent.

Before LLMs, you had spam detection as its own major problem. You had summarization as its own problem. Translation was its own separate problem. There were actually good companies with good ML teams doing each. The thing that they all got wrong was focusing on narrow problems. What we’ve learned is that if you can understand language at the root level, and you see scaling laws, you can get all the downstream applications for free.

So if it is true that there’s a fundamental relationship between geometry and EM fields, just like there is in language, and if scaling laws are true, then this model should generalize.

EM simulation looks a lot like language pre-LLMs. Where Ansys simulators fail – e.g., this tool is for antenna simulation, this tool is for EMI simulation for an automotive engine – we could tackle all of that with one model, like an LLM tackles translation, sentiment analysis, spam detection, and so much more all in one.

We believe that LLMs were the first foundation models, not the last. Language is one primitive of intelligence – the one humans use to communicate and think. But the universe has other primitives. Newton created calculus because the language of the universe is not English. LLMs will let us interface with new foundation models that can push humanity’s understanding of the universe farther, giving us intuition for things our biology wasn’t evolved for. Such models are a critical part of a positive-sum future of AI.

This is the huge bet that we are making at Arena Physica. It’s our thesis. We are already building a strong business that helps companies better understand and iterate on their electromagnetic systems, but we think generalization will also allow us to build the IP Factory for RF.

With our loop, we can automate IP creation and design. This is the key distinction between our model and ARM’s. If generalization works, then instead of selling one design to many customers like ARM does, we can generate a unique design for each customer and each use case almost instantly, with roughly the same effort it takes ARM to create one-size-fits-all IP.

And we think, and are seeing early signs, that our model can scale this automation across the whole electromagnetic spectrum, in which case our TAM is anything with a wave.

One question we get often is: can’t LLMs just do this? We’ve been running our own internal tests against frontier LLMs – both regular and extended thinking - and their performance gap to our base model is substantial. Our model achieves a magnitude weighted-MAE (Mean Absolute Error) well under 1 dB (for context, the range that RF engineers typically care about spans roughly 20-30 dB, so <1dB is a very strong result)6.

What we’re building is different than LLMs, and better at what it was built to do. A useful way to think about what this unlocks is as a “Compiler for Atoms.”

In software, compilers translate high-level programming languages into binary instruction sets that a CPU can execute. We went from assembly to C++ to Python, and now Claude Code is arguably a compiler with English as the programming language. The LLM compiles down to the programming language it decides is best, which then compiles down further to machine code. At each step in this progression, the abstraction gets higher and the number of people who can “program” gets larger.

Physics doesn’t have a compiler yet. The universe has an instruction set: materials and geometries arranged in specific configurations. Place them one way, you get a motor. Place them another, you get an invisibility cloak. We know all of this is possible because the equations tell us so. Human physicists have spent centuries learning this instruction set. But to access any of it, you still need to hire the equivalent of an assembly language programmer: a physicist who has spent decades learning to translate between human intent and the physical world’s instruction set.

What we’re building at Arena Physica is, in some sense, a Compiler for Atoms: a way to express what you want in high-level terms and have it compiled down into the geometries and materials that produce it, starting with Maxwell’s equations and eventually, we hope, adding Schrödinger’s.

When we launch our EM foundation model next week, we’ll make it available to interact with via an agentic UI. You’ll be able to type a request in plain English like, “I need an eight gigahertz band pass filter for a satellite uplink.” The LLM translates that target scattering parameters, the technical parameters we’re optimizing for. Our LFM generates candidate geometries using physics as its reasoning substrate rather than language. Then, the LLM engages again to explain what the model did and why, drawing on the foundational RF engineering knowledge we’ve built into the system.

Andrej Karpathy says LLMs are “people spirits.” In our system, the spirit of David Pozar, author of the bible of RF engineering, provides his best explanation of the geometries generated by our EM foundation model. It’s like working with an intern who happens to speak electron, except the intern is channeling decades of accumulated RF wisdom.

I think this will be a powerful paradigm for the future. Everyone is thinking about human-to-model interactions. But the bulk of the work in systems like ours is model-to-model: the LLM talking to the EM foundation model, the EM foundation model responding, the two iterating through design space at machine speed. The human-to-model layer becomes a minority of the interactions, the intent-setting and interpretation. The real work happens in a language we can’t speak, translated for us by models that can.

In the future, language models might serve as the universal interface between humans and an entire ecosystem of specialized foundation models: for EM fields, for biology, for materials science. In this future, the LLM becomes the diplomat between species of intelligence.

Eventually, we want to get to a world in which anyone can say, “I want an invisibility cloak” or “I want a cheaper Starlink,” and the machine will design it. But today, even with our compiler, we’re still in the C++ era. Expert “physics coders” still need to tell the machine, “I need an 8 gigahertz band pass filter for a satellite uplink.”

In the meantime, we want to deliver that “eventually” future today.

Taking Problems Off Customers’ Plates Entirely

Since it became clear that our model had the potential to scale and generalize, I’ve been thinking a lot about the right way to deliver the capabilities it provides to customers.

I’ve thought about whether we should build the ARM for Analog. I’ve thought about selling access to the model, or the IP Factory, directly. I don’t think either are quite right. I think it’s worth talking through where we landed and how, because what constitutes the right business model is changing a lot with AI, and the model we landed on is probably not the one we would have pursued a few years ago.

If you think through what makes Arena Physica unique, it’s really four things: we have a talent-dense collection of some of the world’s best RF engineers, a “Compiler for Physics” that can generate valuable IP in experts’ hands, a Data Factory that improves as those RF engineers generate and validate more IP, and a software platform (complete with FDEs) that makes it easy to apply LLM-based agents to reason about hardware.

The business model that falls out of that is services. We’ve hired some of the world’s best RF designers and a team of excellent electrical engineers. Each one, working with our EM foundational model, with our agents, and with LLMs, can cover enormous ground. So instead of selling people the tools and asking them to figure it out on their own, we’re starting to just take problems off people’s plates.

You want to launch a space company and you have an unsolved problem with how your racks communicate in orbit? We’ll solve that for you. You need a silicon layout for your next chip? We’ll do the layout. We send in RF and electrical engineers, armed with our tools, and deliver the product the customer needs. It’s full-stack electromagnetic engineering as a service, at a speed and cost that wasn’t possible before, because each of these rare experts can now do what used to require an entire team, faster, better, and cheaper.

And because we’re so early in our journey to build the EM foundation model, working directly with our customers lets us learn faster and improve our models based on real-world needs.

To that end, we’re also pursuing research partnerships. This falls out of our model, too. As we scale the LFM from version one to version two and beyond, we need to decide which training data to generate next, and that depends on the problems we’re solving. Partners working on chip packaging need us to model different structures than partners working on superconducting quantum computing. They get early access to our model and our team for their specific problems, and we get the data we need to drive generalization. The research partnerships feed the Data Factory, the Data Factory feeds the model, and we eat more of the EM spectrum.

One counterintuitive move here is that we don’t plan to sell the model. We plan to publish it, and sell everything around it: the platform, the experts, and solutions to our customers’ problems. The model is what makes all of that possible, but it’s not the product. As Packy wrote in Power in the Age of Intelligence, “If your technology is so good, why aren’t you using it to compete?” Our product is: bring us your electromagnetic problem, and we’ll solve it.

Companies shouldn’t be bottlenecked by the fact that, as it stands, you need hundreds of millions of dollars to build the types of systems we can build. We should be able to almost AWSify expertise for them.

There’s a lot more I want to do here.

As our model improves, and we move from the C++ to Python and even Claude era of EM foundation models, I suspect our business model changes, too. We can sell the “designer” to companies and they can use it to generate their own IP. A sort of Golden Analog Silicon Goose. As that happens, the cost to manipulate the EM spectrum goes down, and humanity’s capabilities increase.

Over the longer term, my dream is to run Arena Physica as a modern Bell Labs, to use the commercial side of the business to fund a new kind of research network.

What if, once we’ve proven the foundation model works, we opened it up? We could give academics free access to our model and our compute. In exchange, when they use it to design novel structures or discover new phenomena, the IP flows back through Arena. We take a cut (maybe 20%, like an app store) and they keep the rest.

Right now, a professor working on some exotic antenna geometry has to write grants, wait for funding, hire grad students, and slowly iterate through simulations on whatever compute they can scrounge. What if instead they could just... use the model? Explore design spaces that would take years to search manually? And get paid when their discoveries become commercially valuable?

If our model can auto-generate IP for any electromagnetic application, we become the platform. The rare humans who can push the boundaries become contributors, and get rewarded for it as their rare knowledge turns scalable. Because the universe of people with this expertise is so small, we can actually get our arms around sharing the upside of their work with them.

Of course, as we expand across the EM spectrum, and our models become multiphysics models, we’re not going to hire all of the world’s great physicists. But by opening up the platform and becoming a new kind of Bell Labs, we can work with them.

I want Arena to be a place where academics can sabbatical in and access our models. Where we’re not just selling software, but funding experiments. Where we create incentive structures that let brilliant people do fundamental research without the soul-crushing grant cycle. Where we compress the cycle between physics research and application, and wield AI not to do what humans can do cheaper, but to do things humans can’t do at all today. If we win, humanity benefits.

But I’m getting ahead of myself. First, we need to take problems off of our customers’ plates.

What Can Our Customers Build?

So what could our customers build if we provide them best-in-class RF and EE?

To start, by working with Arena Physica, any company that wants to do anything with phased arrays can get custom ones, and much more quickly than they could have before.

What they’ll be able to build isn’t limited to what you’d traditionally think of as radar.

I heard a phrase once that stuck with me. I’ve always thought of radar as a detection system, but this guy told me that radar is actually an imaging system. As the frequency gets shorter, you get higher resolution. You can image things. So radar doesn’t just tell you that “something is there.” It also tells you what that something looks like. Radar can create images, like a camera.

Take drones. Everyone is talking about drones as the future of warfare, but currently, we can’t see them with radars or sensors. It might surprise you to learn that the United States Navy doesn’t have counter-drone phased array radar at scale. That surprised me too, so when we dug a little bit, what we heard was that in the current system, it’s just too expensive. To get Raytheon to build a new ship-based phased array radar for them would end up costing as much as half the ship.

It should not cost close to a billion dollars to make phased array radars. But remember that slow process of imagining and simulating designs that we discussed earlier? Now, imagine that happening inside of a slow-moving legacy prime that gets paid a margin on top of every dollar it spends. The result is that our Navy doesn’t have phased array radars that are good at detecting drones.

We can help the Navy see drones much more cheaply. And we should. For drone detection, you might have 1,000 incoming targets. The benefit of phased arrays is that you can form multiple beams from one antenna, like Starlink does, instead of something spinning like an old-timey radar or LiDAR. (As a side-note, those spinning LiDARs on top of Waymos will almost certainly become phased arrays, too, and when they do, the whole unit goes solid-state, which is cheaper and more reliable with no moving parts).

We also have people reaching out to us who want to design phased arrays for drone capture. They’re building systems that catch drones in mid-air with a robot arm. The drone capture systems are smart; they have onboard computing and sensors to actively track and intercept the incoming drone. They need to be able to see, which means they need custom phased arrays that can track a fast-moving drone with high precision, work in any visual conditions, and are small and cheap enough to put on the capture mechanism itself. They certainly couldn’t make the math work at a billion dollars, but we can bring those costs down at least an order of magnitude in the near-term with automated design and fast iteration.

The applications only multiply once you realize that you’re dealing with an all-condition imaging system, and that drones are just one application of phased arrays. The same physics that steers radar beams also steers communication beams.

If space stuff continues to grow the way everyone thinks it will, you’re talking about tens of thousands of satellites and millions of ground terminals. Every single one needs these precisely shaped silicon tiles. Every ground station needs phased arrays. Northwood just raised a ton of money to build phased array ground stations. Every Starlink antenna on a house and satellite is powered by phased arrays.

Now, let’s say an adversary is trying to jam your radar signal. Notice how these little things dive down to zero:

Remember I said you can dynamically move this beam? This is one of the benefits of the phased array. It lets you move those points. Those are called null points, and it’s where the interference pattern zeroes out, like your noise canceling headphones. So imagine I’m being jammed. What’s crazy cool about a phase array is that I can transmit my signal, and then I can cyclically move my null point to absorb the jam signal. Just think about it. It’s remarkable, actually. You’re still transmitting and receiving, but you’ve carved out a little pocket of silence right where the enemy is screaming at you.

But if you can exploit the physics really well—like, in this case, we’re still using our good old digital transistors to say what to do, while the analog shapes determine the quality of the EM fields produced—this is where Gordon Moore’s dream might come true: his silicon boolean transistors are talking to analog silicon. And that’s really cool. With digital silicon, we can compute. But with analog silicon, we can transmit power, we can transmit directed energy, we can absorb energy. Analog silicon makes it much more physical.

The same physics that lets you communicate also lets you deny communication to others. And the same physics that lets you transmit also lets you absorb. Stealth is just the inverse problem of radar: instead of bouncing signals back, you’re making them disappear.

It’s all shapes.

Imagine we could change the economic structure of all of this.

That is exactly what we are trying to do at Arena Physica. Our mission is to create “Electromagnetic Superintelligence.” It sounds audacious, but remember, it is much easier to achieve superintelligence—as in, relative to humans—in electromagnetism than it is in language or even math. And it describes precisely what we’re building: a system that develops superhuman intuition for how geometry shapes electromagnetic fields, a mind that can see what we can’t.

As a software engineer, I don’t know how to be an infrastructure engineer. That’s because I don’t need to. Amazon takes care of it for me. If you could say companies no longer need RF expertise, you could lower the cost of everything we’ve discussed dramatically by a factor of 10 or more. We could give RF capabilities to everyone, from small companies to those that serve the Navy.

One obvious ramification is we’ll probably see more satellite companies and space companies, because they can now design their own phased arrays. More competition in the radar space. More competition in the jamming space.

Another less obvious one might be backpack radars. Think about troops going into a situation like Ukraine. One of the big risks they face is drones sneaking up on them. They should have backpack-mounted counter-drone radar: small, cheap phased arrays that let every warfighter see what’s coming.

Backpack radars are a very specific thing, but the point is that they’re something that made no practical or economic sense before that becomes practically and economically feasible.

We can even help AI models get better in a sort of indirect way.

Data centers need to move enormous amounts of data between chips very fast. The problem is that, at those speeds, the wires connecting chips start acting like antennas. They can accidentally broadcast and pick up signals. Which means they hit bandwidth limits on chip-to-chip communication. So their goal is the opposite as ours: they’re trying to make really bad antennas. They don’t want the electron traveling between their GPU and CPU to pick up a signal or transmit one, because then the data gets garbled. This is called signal integrity. The solution is the same shaped-silicon approach: carefully designed structures that guide high-frequency signals without interference.

It’s not just chip-to-chip. Think bigger. A data center company recently asked us whether we could beam data rack-to-rack wirelessly, because the cabling itself is becoming a bottleneck to how fast they can deploy. It’s not just terrestrial, either. For orbital data centers, there won’t be any option. You’re not going to be running optical cables between racks in space.

The most interesting thing is how the market has responded by asking us for use cases we would never have thought of.

For example, high-frequency trading firms have reached out to us about helping them trade faster. I’d assumed fiber optics already transmitted at the speed of light, but due to total internal reflection inside the glass, signals travel at only 60-70% of light speed. A phased array transmitting through free space goes at actual light speed. Over the distance between New York and Chicago, that difference could be enough to make a lot of money.

If you look at what we’ve just described, there are really two different things happening. I think we‘re going to see a K-shaped future for hardware.

The lower leg of the K is making commodity devices dramatically cheaper and more accessible. For example: the Navy acquiring drone-detection radar without writing a billion-dollar check to Raytheon, satellite startups designing their own phased arrays instead of outsourcing to a prime, backpack-mounted counter-drone radar for every warfighter, and data centers deploying faster with wireless rack-to-rack links are capabilities that exist today in expensive, bottlenecked forms. We want to remove these bottlenecks and make the capabilities cheaper and more abundant.

The upper leg of the K is making exquisite devices at the frontier of what’s physically possible newly achievable. This is what Bell Labs enabled, entirely new and more powerful capabilities than were possible before their research. There is a lot of excitement about cheap, attritable systems in the future of warfare, and rightly so. But in the conversations I’ve had with military leadership, they believe that to win (or deter) a conflict in the Indo Pacific, we’re going to need some of the world’s most exquisite machinery. The F-117 was crucial to winning the Gulf War; it made up just 2.5% of coalition air power but destroyed 40% of all strategic targets. We want to make new types of exquisite hardware possible, for defense and beyond.

By making it easier, faster, and cheaper to design more capable RF components, we think we’ll help expand the market beyond current analyst estimates, which don’t anticipate the unimagined. Those estimates project a 12% CAGR for RF components, from $44.8 billion to $140.5 billion over the next decade. I think that’s wrong, almost certainly too low. One of the reasons they’re projecting relatively slow growth is that RF is just too hard today. But if you democratize the expertise and transform the cost structure, would everyone just add better all-weather sensing equipment onto their robot? Would everyone just add backpack-mounted radars to every soldier?

What I’m most excited about is this open possibility space. I don’t even know what else people might think up now that they have the ability to manipulate this stuff. If we’re right, their ambitions won’t be limited by speed, economics, or even by the shapes required to deliver the capabilities that they need.

Alien Designs

A lot of the shapes coming out of our system already look nothing like what a human would design. Basically, we’re working with near-alien humans to create a system that will ultimately produce alien designs.

Human RF engineers have been trained on certain canonical structures: dipoles, patches, spirals, horns. They know these shapes work because they’ve been refined over decades. When they design something new, they start from these familiar forms and tweak them.

Our system doesn’t care about any of that. It starts from noise and evolves toward function. The results often look like QR codes, or random stippling, or structures that seem to follow no logic at all.

When we show these designs to expert RF engineers, their first reaction is usually skepticism. “That doesn’t look like an antenna.” “I would never have come up with that.” “Are you sure this works?”

Then, we fabricate it.

Today, we’re fabricating at the PCB board level. The goal I have for the team is that we’ll do our first silicon tapeout this year. As in, we’ll manufacture actual silicon chips. Analog silicon has the advantage that we don’t need TSMC’s bleeding-edge fabs; older, cheaper factories like Samsung, Global Foundries, and some defense fabs can do it, because the node size is typically larger.

And it works. This is the AlphaGo moment. Remember Move 37 expert commentary?

We’re seeing the same pattern. Engineers look at our designs and say, “That’s super unconventional. That’s unusual. That’s not in the textbook.” And then it works. And then they say it was “creative.”

Our goal is to do more than just match human performance autonomously. We want to exceed it and find all of the EM equivalents of Move 37: designs so counterintuitive that no human would have tried them, but so effective that they outperform anything we would have tried.

My hunch is that this will happen fairly quickly in EM because of what we discussed at the top: most humans didn’t evolve to intuit which shapes produce which EM waves. Therefore, we are terrible at intuiting the answer. On the other hand, computers can become superhuman — we can evolve them to get there in simulation.

There’s this great story in Skunk Works, Ben Rich’s personal memoir from his time atop the famed Lockheed division, about how the F-117 stealth bomber came to be.

A thirty-six-year-old mathematician and radar specialist named Denys Overholser happened to read a translation of a dense technical paper by Pyotr Ufimtsev, the chief scientist at the Moscow Institute of Radio Engineering, titled Method of Edge Waves in the Physical Theory of Diffraction. Ufimtsev had “revisited a century-old set of formulas derived by Scottish physicist James Clerk Maxwell… these calculations predicted the manner in which a given geometric configuration would reflect electromagnetic radiation” and took them a step further. “Ben,” Overholser told Ben Rich, “this guy has shown us how to accurately calculate radar cross sections across the surface of the wing and at the edge of the wing, and put together these calculations for an accurate total.”

With Ufimtsev’s work, the Skunk Works team could create computer software to calculate the radar cross section (how visible an object is to radar) as long as the shapes were in two dimensions. If they designed the bomber as thousands of flat triangles, they could add them all up and get the radar cross section.

That’s exactly what Overholser did, and the design that emerged “was a diamond beveled in four directions, creating in essence four triangles,” which, viewed from above, “closely resembled an Indian Arrowhead.”

He called it the Hopeless Diamond, and calculated that it would be “one thousand times less visible than the least visible shape previously produced at the Skunk Works.” On a radar screen, it would appear to be the size of an eagle’s eyeball.

Kelly Johnson is the Skunk Works founder and boss who was so magnificent at airplane design that his boss said of him, “That damn Swede can actually see air.”

He was so unimpressed by the design that, when he saw it, he physically kicked Rich in the butt, crumpled the proposal, threw it at Rich’s feet, and stormed, “Ben Rich, you dumb shit. Have you lost your goddamned mind? This crap will never get off the ground.”

As it turned out, the model was right and even the great Kelly Johnson was wrong. The Hopeless Diamond became the F-117 Nighthawk, the stealthiest plane built to date by more than three orders of magnitude, and flew over 1,300 sorties during the 1991 Gulf War without a single combat loss.

The F-117’s was an alien geometry, and it gave the US otherworldly capabilities.

We’re attempting to do something similar, but at silicon scale (to start). And with AI doing the search instead of human engineers sketching on whiteboards, we’re planning to do it over much wider problem spaces and search spaces than currently exist.

The Shape of Things to Come

We have a lot of work ahead of us to solve the core problem: building a Large Field Model that can do for EM what LLMs did for language.

The simulation has to get faster. The generative model has to get smarter. The fabrication loop has to tighten, and we need to actually fab analog silicon, and then we need to do it for a lot of customers. We need to hire more of those rare humans who think like electrons and work with them to create training data.

It’s hard to see what our models can do today and not imagine what they might be able to do in the future, though.

If it scales across the entire EM spectrum like we think it will, things will get very interesting.

How interesting? Packy asked me whether our models might one day help finally produce a Grand Unified Theory, if one is possible.

I don’t know. But here’s how I think about discovering new physics more broadly.

All of our tools today — since the dawn of computers — have been about deduction. Solve, compute, calculate, predict. Here’s an input; tell me what happens. But a lot of really creative human reasoning is inductive. You postulate a question or an idea, and then you research it. That’s deeply creative and deeply human.

This connects to something personal for me. My undergraduate research advisor at Stanford, Hari Manoharan, did an experiment that made the cover of Nature in 2000.

He arranged 80 cobalt atoms in an ellipse on a copper surface. You know those whispering galleries where you stand in a corner and whisper and someone on the other side can hear you? That’s constructive interference of sound waves. Hari knew that electrons also behave like waves, and he suspected they should interfere in the same way.

And that’s exactly what he saw. A “quantum mirage.” A ghost atom appearing at one focus when a real atom sat at the other. What blew everyone’s mind was that there was no time delay. Physics predicted some tiny delay to account for information traveling at the speed of light. Instead, it was instantaneous. This kicked off a whole field of quantum communication research.

Hari knew the Schrödinger equation. He knew Maxwell’s equations. Everybody knew those equations. But he combined them in a way no one had thought to try, postulated what might happen, and built an experiment to test it. That’s how new physics happens.

How amazing would it be if we had machines that could actually understand these equations and help us? Currently, we have to wait decades for a genius to come along and push a field forward. What could we benefit from if those geniuses had a little help? How much closer could we pull the future? How much more of the universe could we understand in our lifetimes?

What if our foundation model could start making those leaps? Right now, it’s not yet breakthrough inductive, meaning it can’t create its own experiments. It’s like an applied physicist: we can tell it our engineering goal and it will come up with a design. But as it learns more, develops something like intuition, could the model start postulating? Could it notice patterns humans have missed and suggest experiments?

And maybe (this is where I let myself dream), maybe if we build foundation models for each of the fundamental forces, and they start talking to each other, we get closer to something bigger. There are four fundamental forces: strong, weak, gravity, electromagnetism. We’re just trying to make a dent in one of them. But physics is deeply interconnected. Hari’s quantum mirage happened because electromagnetism and quantum mechanics intersected in a way that nobody expected.

What happens when foundation models that understand multiple forces chain together, exploring the spaces between them? I can’t stop thinking about this.

Ultimately, that’s what this is about. It’s why I pursued my PhD in quantum electromagneticism for four years (then dropped out in true Silicon Valley fashion) and why I started Arena Physica.

I want to understand the nature of reality in order to manipulate it for the betterment of humanity. To do that, we need models that learn directly from physics, that develop intuitions we’ve never evolved to have.

One of the questions that came up in grad school, and I remember thinking, how do people even ask these questions, was: Why does physics work? Why does math work? Isn’t it strange that reality is so... describable? Why isn’t it much more random?

Maybe these models will help us find out.

Electromagnetism secretly runs the world. We’ve been manipulating it for a century with our hands tied behind our backs, limited by the rarity of humans who can see what we cannot see.

We’re creating something to understand the interplay between geometries and electromagnetic waves, and evolving them to develop a new intuition.

The universe is made of fields. Fields are shaped by geometry. Geometry, it turns out, is something computers can learn much better than we can. We should lean on them so that we can get to new problems.

If we learn to shape the waves, we might be able to shape the future.

Big thanks to Pratap and the whole Arena Physica team for sharing their knowledge, and to Badal for the cover art.

That’s all for today. We’ll be back in your inbox with a Weekly Dose on Friday.

Thanks for reading,

Packy

Hi friends 👋 ,

Happy Friday and welcome back to our 185th Weekly Dose of Optimism. Fresh off the heels of my World Models primer with General Intuition’s Pim de Witte…

… I’m excited to bring you my favorite Dose in a long time. Kalanick, Bezos, copper, cancer drugs, meditation stimulation, and the return of Ulkar. I can’t wait any longer…

Let’s get to it.

Today’s Weekly Dose is brought to you by… Framer

Framer gives designers superpowers.

Framer is the design-first, no-code website builder that lets anyone ship a production-ready site in minutes. Whether you’re starting with a template or a blank canvas, Framer gives you total creative control with no coding required. Add animations, localize with one click, and collaborate in real-time with your whole team. You can even A/B test and track clicks with built-in analytics.

The weekend starts today, and the weekend is the perfect time for side-projects. Get to it.

Launch for free at Framer dot com. Use code NOTBORING for a free month on Framer Pro.

Just Publish it With Framer

(1) Travis Kalanick is Back with Atoms

Travis Kalanick &

There’s a version of the story where Travis Kalanick disappears into the wilderness after his investors ousted him from Uber. He’s rich. He doesn’t need to deal with this shit.

If you watched the Uber saga, though, you knew that wasn’t how the story was going to play out, and luckily, you were right. Travis. Kalanick. Is. Back.

Kalanick spent nearly eight years in what might be the most extreme version of stealth mode any modern founder has pulled off: he’s hired thousands of employees, in 30 countries, bought and developed hard real estate assets, built a full-stack food infrastructure business, and did it all like lasagna. Employees weren’t even allowed to list the company’s name anywhere.

The company was called City Storage Systems. Its most visible subsidiary, CloudKitchens, operated ghost kitchens. It was valued at $15 billion in 2022. And it was just, to use a food term, a little amuse-bouche before the main course.

Last week, Kalanick unveiled Atoms: a robotics company spanning food, mining, and transport, built on everything CloudKitchens learned about physical automation. CloudKitchens is now Atoms Food, which includes Lab37's Bowl Builder robot (200 meals per hour, no humans), the Otter restaurant OS, and Picnic delivery.

With Atoms, TK is expanding into mining (via the acquisition of Pronto AI, an autonomous haulage startup for mines and quarries founded by Waymo founder Anthony Levandowski) and transport, where Atoms is building what it calls “a wheelbase for robots.”

Kalanick’s robotics bet is explicitly anti-humanoid. It’s the same bet that we wrote about in Many Small Steps for Robots, One Giant Leap for Mankind with Standard Bots’ Evan Beard and in yesterday’s World Models: Computing the Uncomputable with General Intuition’s Pim De Witte. “The recent humanoid Olympics in Beijing highlighted many advances in humanoid development,” Kalanick writes in the company’s Vision doc, “I watched the half-marathon and couldn’t help but think how much better it would be if they just had wheels.”

Uber was Kalanick’s first attempt to digitize the physical world. Atoms seems to be the rest of it.

It’s a bad idea to bet against Travis Kalanick. Watch the TBPN interview to get a taste of why. Which means it’s a good idea to bet on our autonomous everything future.

(2) Jeff Bezos Raising $100B AI Manufacturing Fund

The Wall Street Journal

THE BOYS ARE BACK IN TOWN. First Kalanick, now Bezos. Elon’s been living here.

Imagine going out to raise one of the largest funds in history, a SoftBank Vision Fund-sized whopper of an investment vehicle, one that exceeds the total amount of US venture capital funds raised in 2025… and that is less than 50% of your personal net worth. Jeff Bezos doesn’t have to imagine.

The WSJ reported that Bezos is looking to raise a $100B fund “to buy companies in major industrial sectors such as chipmaking, defense and aerospace.” The plan is to buy existing manufacturing companies, and then use Bezos’ new startup, Project Prometheus, which sounds like a World Models company (“building AI models that can understand and simulate the physical world”), to boost efficiency and profitability.

Basically, the fund is to AI rollup funds targeting accounting firms as new Jeff Bezos is to scrawny old Jeff Bezos. It’s the gigachad version of the AI rollup playbook.

Details are sparse, but if you told me a couple of years ago, in the depths of the bear market, when everyone said that America couldn’t build things anymore, that Jeff Bezos, Travis Kalanick, and Elon Musk were going to be competing to industrialize America harder, all while Chris Power launches a new Hadrian factory in Alabama, I would have told you… welcome to The American Millennium.

(3) Copper One: The World’s Only Autonomy-First Mine and Refinery

Mariana Minerals

Wait… and we’re mining and refining autonomously here, too?

Long time not boring readers will know that I am a big fan of mining startups, given how critical (get it) the industry is to the modern world and how critically underinvested-in it’s been. I think sell side analysts will write poems about Earth AI’s business model one day.

It’s also a unique industry in that it’s structurally not winner-take-all. Earth AI can build a very, very big business while Kobold, Durin, and Mariana Minerals do, too. If you have enough high-grade metal in your deposit and can get to it economically, you’ll make money.

So I was pumped to see Mariana announce Copper One, its first autonomy-first copper mine and refinery in Utah. The blog post in which they announced the project is one of the best company announcement blog post’s I’ve read in a while, with clear explanations and great graphics, so I encourage you to go read it. Here, I’ll just share the plan they laid out to turbocharge the existing copper mine they acquired last year

1. Deploy PlantOS at scale to maximize copper recovery and reduce copper refining costs throughout heap leaching, solvent extraction, and electrowinning.

2. Restart mining operations with autonomous equipment and orchestration via MineOS.

3. Integrate copper scrap processing into the refining circuit, leveraging PlantOS to manage feedstock variability and to put a meaningful dent in US copper scrap exports.

4. Scale combined output at the site to 50,000 metric tonnes per year from both geologic and scrap feedstocks (leveraging CapitalProjectOS to accelerate capital project delivery).

We’re going to need to find a lot of metals and minerals, we’re going to need to get them out of the ground and refine them more efficiently, quickly, and cleanly. I’m excited to see what improvements Mariana can dig up in Utah.

(4) Scientists Create Cancer-Fighting Immune Cells Right in the Body

University of California San Francisco

Long time not boring readers will also know that we have long taken a bold but important staunchly anti-cancer stance here in the Dose. We just don’t like cancer, and we want to see it gone.

’s piece on Sid Sijbrandij’s extraordinary care journey, Going Founder Mode on Cancer, remains our favorite in the anti-cancer canon. It tells the story of one person’s against-the-odds battle to bend the medical system and cure his own cancer. But it shouldn’t have to be that hard. Enter this week’s entry.

CAR-T therapy is one of the most powerful weapons against cancer. It works by pulling a patient's T cells out of their body, genetically reprogramming them to hunt cancer, growing them up in a lab, and infusing them back in. Seven CAR-T therapies are now FDA-approved for blood cancers. The problem is, the process takes weeks, costs over $400,000, requires specialized manufacturing facilities, and demands lymphodepleting chemotherapy just to make room for the new cells. Most cancer patients in the world will never have access to it.

This week, a team at UCSF led by Justin Eyquem published a paper in Nature showing they can skip almost all of that. Instead of the extract-engineer-expand-reinfuse pipeline, they designed a two-particle injection system that reprograms T cells inside the body. One particle delivers CRISPR-Cas9 to make a precise cut in the T cell’s genome. The second delivers the DNA template for a chimeric antigen receptor (the cancer-targeting weapon). The whole thing is designed so only T cells get edited, and only at a specific, safe genomic location, avoiding the random integration that can, in rare cases, cause secondary cancers.

In mice with humanized immune systems, a single injection cleared detectable leukemia in nearly all animals within two weeks. The engineered cells made up as much as 40% of T cells in organs like bone marrow and spleen. The approach also worked against multiple myeloma and sarcoma, a solid tumor that is historically much harder for CAR-T to crack. Best of all, the in vivo cells actually outperformed lab-manufactured ones, because cells that never leave the body retain their “stemness” and ability to keep dividing.

As always, it’s important to remember that mice are not people, and we need to see this thing work in humans before we start popping the champagne. To that end, Eyquem and his collaborators founded a company, Azalea Therapeutics, to push toward human trials. If it translates, this could turn CAR-T from a last-resort therapy available at a handful of elite cancer centers into something closer to a vaccine, a single injection that any hospital could administer.

Just a quick shot, a follow-up, and back at it. Say it with us… get fucked, cancer.

(5) Facilitating Mindfulness Training with Ultrasonic Neuromodulation

Brian Lord, Erica N. Lord, Jessica Schachtner, Laura Beaman, Shinzen Young, John J. B. Allen, Joseph L. Sanguinetti

A little while back, I read Shinzen Young’s The Science of Enlightenment. It was one of the best books I’ve read on meditation. Shinzen is a Jewish-American who trained as a Shingon monk in Japan, did deep practice in all three major Buddhist traditions (Vajrayana, Zen, Vipassana), and then came back to the West to fuse contemplative practice with scientific rigor.

Most relevantly, near the end of the book, he makes a wild prediction about Maitreya, the prophesied “future Buddha” of Buddhist tradition. Maitreya won’t be a person, Shinzen argues. It will be a collective of scientists, technologists, and contemplatives working together to make liberation accessible at scale through tools and systematic methods, rather than requiring every individual to spend decades in a monastery.

This week, Shinzen and a team of researchers dropped a preprint on bioRxiv that might be a step in that direction.

After dozens or hundreds of hours of sitting, experienced meditators have brains that look different from everyone else's. Their default mode network (DMN, the brain's self-referential chatter machine, the thing running rumination loops, and the thing that psychedelics seem to quiet) decouples from the central executive network, the system that handles focused attention. That decoupling is the neural signature of equanimity: letting experiences arise and pass without getting caught up in them. It's associated with reduced stress, lower depression, and generally being a calmer human. But it takes hundreds of hours to get there, and most people quit long before anything rewires, because the early stages are so uncertain and even boring.

What if you could use technology to do the same thing, much faster?

Shinzen and the team at the University of Arizona ran a randomized controlled trial with 24 meditation-naïve participants who did a two-week mindfulness program. Half received transcranial focused ultrasound targeting the posterior cingulate cortex (the hub of the DMN) during four in-person sessions. Half got sham stimulation.

After two weeks, the active group's brains showed the decoupled network pattern that normally takes experienced meditators hundreds of hours to develop (p < 0.001). The sham group's connectivity actually increased; they got more tangled up, not less, which is exactly what frustrated novices do. Within the active group, greater decoupling also predicted bigger increases in self-reported acceptance and longer voluntary meditation sessions. The ultrasound made the practice better and gave novices experienced meditator brains.

It's a small study and a preprint, so it needs replication, but as someone who has spent a lot of hours trying to go deep, it also seems remarkable.

I love this because I’ve written that the worst outcome would be for us to get all of the technological wonders we could have asked for and still be unhappy, and I think meditation, and the associated ability to pay attention, can help there.

It’s also just very cool to see someone who spent his entire career arguing that the cross-fertilization of Eastern contemplative technology and Western science would eventually produce something neither could produce alone being proven right.

“The next Buddha is a sangha,” indeed.

EXTRA DOSE:

1. Scientific Breakthroughs with Ulkar Aghayeva

2. My Favorite Essay of the Week

Read more

Subscribe now

Hi friends 👋 ,

Happy Wednesday!

A few months ago, Pim De Witte and Kent Rollins invited me to their office right here in New York City to show me what they’ve been cooking up at General Intuition. I’d heard about the company, from the announcement of their leet $133.7 million Seed round, and I’d heard about the class of product they were building, World Models, but I didn’t know much beyond that.

What they showed me that day, models that learn to predict the near future from action-labeled gaming clips, and what I’ve learned from many conversations and dozens of hours of research since, has changed my perception of what models can do. I am on the record as being skeptical that LLMs will take us to superintelligence, but I think there is a real shot that World Models will drive superhuman, complementary machines that do things that we can’t, or don’t want to, do.

Since that first meeting, the World Models space has heated up. Fei-Fei Li’s World Labs raised $1 billion. Yann LeCun’s AMI raised $1.03 billion. World Models were one of the stars of this week’s NVIDIA GTC. But the field is so nascent and there is so much going on, so many geniuses pursuing competing and collaborative approaches, that it’s hard to make sense of it all.

So I asked Pim to team up with me on a co-written essay about the history, theory, progress, and potential of World Models. He agreed, and both he and the General Intuition team have been incredibly generous with their time and human intelligence in helping me get up to speed, so that I can help you get up to speed.

I have the coolest job in the world. Over the past couple of months, I’ve gotten a front row seat to the future of embodied AI, of Models and Agents, trained in dreams, that direct machines to do things for us in the physical world.

I’m thrilled to share the fruit of that exploration, what I think is the most comprehensive guide to World Models that exists. Obviously, Pim and the GI team have a perspective on the best way to build World Models, but I was impressed with how careful they were to present the pros and cons to every approach, including theirs, and with their admission that the future is not yet determined.

The space continues to change and progress incredibly fast. I hope this will help you navigate and make sense of all of the exciting news that continues to drop.

Let’s get to it.

Today’s Not Boring is brought to you by… Framer

Framer gives designers superpowers.

Framer is the design-first, no-code website builder that lets anyone ship a production-ready site in minutes. Whether you’re starting with a template or a blank canvas, Framer gives you total creative control with no coding required. Add animations, localize with one click, and collaborate in real-time with your whole team. You can even A/B test and track clicks with built-in analytics.

Launch for free at Framer dot com. Use code NOTBORING for a free month on Framer Pro.

Just Publish it With Framer

World Models: Computing the Uncomputable

A Co-Written Essay with Pim De Witte

“I wanted to fall asleep last night. Instead, I started imagining all of the scenarios I might run into the next day, and how I might react to them.”

This is a common experience. As humans, we imagine easily, whether it’s complex sports stadiums, potential romance, or heated discussions. We don’t have to work harder to imagine ourselves at the next Manchester United game than we do to imagine talking to a friend we’ve known for years, even though imagining a Manchester game includes simulating and modeling the behavior of thousands of people, something that would take years for traditional computers and game engines today1.

Think about writing the code to describe the Man U match: at any moment, a fan might bring a random, home-crafted flag. The entire stadium starts singing a song related to it. Only some will sing, though; others will jump with their kids, while an old couple sits still, wondering if this is their last game together, soaking in every second in silence.

The world is a place where unexpected futures unfold, but in somewhat predictable ways. As humans, we can envision almost all of them with roughly the same amount of effort with a very similar amount of time given to each thought. Computers can’t.

It’s no wonder traditional computing struggles with this complexity. Imagine anticipating and coding each and every action, as well as the interactions between all of those actions. Mathematically, in a traditional engine, simulating N fans is at least an O(N) or O(N²) problem. Each person, flag, chair, and ball must be explicitly calculated — and really, the interactions between them need to be calculated, too.

In robotics, machines must respond to situations in the real world in the same amount of time, regardless of their complexity, even though, in traditional computing, different situations can take wildly different amounts of time to simulate. This has been a major bottleneck for robotics and embodied AI progress.

World Models are a solution to that problem.

World Models learn to predict those dynamics from video and, often, the actions taken in them. They reduce situations that are dynamic and computationally difficult to simulate at scale — including stochastic, action-dependent group behavior like soccer games — into a single fixed cost operation in a neural network.

In a World Model, the entire stadium is simulated as a fixed cost forward pass through the neural network. The complexity of the scene doesn’t exponentially slow down the ‘engine’ during inference because the weights have already absorbed the patterns of the world in training.

How? Actions.

Actions act as a form of compression to predict unfolding dynamics: they hold the information to unroll future states in an environment, until more actions take place and add new inputs into the environment. Each action carries enough information to predict what happens next, until the next action updates the picture.

This action-conditioned approach allows models to learn and plan interactively. Today, this is intractable in even the best simulation engines, and definitely not at predictable compute costs. Actions help models interact with the world like we do.

Over and over again, every single day, you observe, you compute, you decide what to do, you act. This is life. At any point, all gathered information about space and time collapses into the action you take.

For computers, actions are a cheat code around the costs of simulation. If human brains are much more efficient than best-in-class LLMs, then we can get all of that computation practically for free by observing how humans respond to the countless variables in their environments. This gives us a way to do non-deterministic computing efficiently and create simulations that shouldn’t be possible under traditional compute constraints.

This ability to compute the uncomputable is why we believe World Models will unlock progress in embodied AI in a way that current model architectures can’t.

Think about models like dreams.

Have you ever had a dream where you simply stood and watched what was happening without the ability to intervene? That’s a video model.

The real world is different. It responds to what you do or instruct to do, and predicts the full range of things that could happen as a result, not just the single most likely or most entertaining next frame.

Have you ever had a lucid dream in which you were able to shape the story inside the mind-generated dreamscape? That’s a World Model.

I coded up a comparison that you can play with here.

More formally, while a standard video model predicts the next frame based on probability, P(x_t+1 | x_t), a World Model predicts the next state based on intervention, P(s_t+1 | s_t, a_t).

That a_t, the action at time t, is the magic.

At General Intuition, we believe (and are seeing early signs) that World Models are a new and potentially more powerful class of foundation model than LLMs for environments that require deep spatial and temporal reasoning. Environments like our real world.

World models — these systems that learn from watching the world and the actions taken in it — are a fundamentally new kind of foundation model. They can compute what was previously uncomputable.

They will matter far more than anyone currently realizes, because they offer a path to general intelligence that language and code alone cannot. Being human, after all, is spending a lifetime taking actions based on what we experience, observe, and learn.

Pause. You might be confused by that claim, that World Models offer a path to general intelligence that LLMs cannot. Understandably so.

World Models are getting a lot of attention as of late. Yann LeCun, who has been skeptical that LLMs are the path to general intelligence, just announced that he raised $1.03 billion for AMI. Fei-Fei Li’s World Labs has also raised more than $1 billion to pursue World Models. Google DeepMind, which has the closest thing to an infinite money printer in tech, is betting money on World Models too. But what we’ve seen so far from that investment are cool videos and 3D worlds.

LLMs can quote Shakespeare and solve Erdős Problems. World Models, on the other hand, still seem more like a path to the Metaverse than a path to general intelligence.

But part of the reason World Models don’t yet have the hype of LLMs is that their definitions are still shaky.

What are World Models? We’ve already said that video models don’t fit the definition. 3D space models don’t, either. That said, both may be paths to World Models. Are the models that animate robots today World Models? Not really, although some are, and even the ones that aren’t share features with World Model architectures.

As always, hype adds to confusion. “My prediction is that ‘World Models’ will be the next buzzword,” Alexandre LeBrun, the CEO of AMI Labs (which is definitely a World Model company) told TechCrunch. “In six months, every company will call itself a World Model to raise funding.”

Hype is a small part of it. What we — and everyone else building in this space — believe is that World Models are the path to controlling machines in the physical world. There are differences in what we believe this path will look like. But all of us believe that the future runs through World Models.

“...very few understand how far-reaching this shift is…,” NVIDIA Director of Robotics and Distinguished Scientist Jim Fan said recently. “Unfortunately, the most hyped use case of World Models right now is AI video slop (and coming up, game slop). I bet with full confidence that 2026 will mark the first year that Large World Models lay real foundations for robotics, and for multimodal AI more broadly.”

Today, we’d like to welcome you into the group of the “very few” who “understand how far-reaching this shift is.” We are going to share the history of World Models, the state of the field as it stands today, broad explanations of the approaches each major lab is taking, and the convictions that drive General Intuition’s directions.

Whether you come with us is up to you. You take the blue pill, the story ends. You wake up in your bed and believe whatever you want to believe. You take the red pill... you stay in Wonderland, and we show you how deep the rabbit hole goes.

For example…. how can you be sure that you’re not an Agent operating inside of a World Model yourself?

Can Agents Learn Inside of Their Own Dreams?

Wake up, Neo.

World models aren’t a new idea. They are one of our oldest. Since humans gained the ability to think about our place in the universe, to ask why we are here, we have pondered whether our reality is just a simulation.

In 380 BC, Plato, via Socrates, offered The Allegory of the Cave. Imagine human beings who live underground in a cave, necks chained, forced to look ahead at the shadows on the wall. Those humans would believe those shadows to be reality, when in fact they are mere shadows of reality. This was Plato’s metaphor. He suggests that we are all stuck in the cave, necks chained, mistaking our perception for true reality.

Eighty years later, Chinese Daoist philosopher Zhuangzi contemplated similar questions in a passage of his Butterfly Dream:

Once Zhuang Zhou dreamt he was a butterfly, a butterfly flitting and fluttering around, happy with himself and doing as he pleased. He didn’t know he was Zhuang Zhou. Suddenly, he woke up and there he was, solid and unmistakable Zhuang Zhou. But he didn’t know if he was Zhuang Zhou who had dreamt he was a butterfly, or a butterfly dreaming he was Zhuang Zhou. Between Zhuang Zhou and a butterfly there must be some distinction! This is called the Transformation of Things.

As the centuries passed and our technological capabilities evolved, sci-fi writers joined the long lineage of thinkers inquiring about the true nature of reality. Frederik Pohl’s 1955 The Tunnel Under the World. Daniel F. Galouye’s Simulacron-3. Stanislaw Lem’s Non Serviam. Vernor Vinge’s True Names. William Gibson’s Neuromancer. Neal Stephenson’s Snow Crash. All painted textual pictures of simulated worlds.

During a 1977 speech in Metz, France, sci-fi legend Philip K. Dick confidently told the audience: “We are living in a computer-programmed reality, and the only clue we have to it is when some variable is changed2, and some alteration in our reality occurs.”

Your first interaction with the simulation was probably The Matrix. Ours was. In the original script for The Matrix, the Wachowskis conceived of the Matrix as a simulation collectively produced by human brains chained into a neural network.

The studio thought humans-as-computers was too confusing a concept for mass-market audiences, so they made the thermodynamically questionable decision to turn humans into batteries that powered the simulation. That was probably the right commercial call. The Matrix franchise has done nearly $2 billion in worldwide gross. More impactfully, it introduced the masses to the idea of a simulated world generated indistinguishable from the “real” one.

It’s no wonder that this idea has taken hold of our collective imagination. It’s certainly the right kind of weird but it’s also surprisingly hard to disprove. If the observations are the same, and the actions are the same, then the computation is the same. If what you see is the same and what you do is the same, it doesn’t matter whether you’re in a simulation or reality. It doesn’t matter whether you’re walking down a real street or a simulated one. Your brain processes both identically. Neo had no idea he was in the Matrix until Morpheus woke him up.

Christopher Nolan, throwing audience confusion to the wind — savoring it, even — released Inception3 in 2010. Dreams within dreams within dreams.

Nolan’s central premise is that the dream is a controllable space from which information can be extracted or, more importantly, into which information can be implanted.

But it’s all just sci-fi, right?

In 1990, Jürgen Schmidhuber, a young researcher at the Technical University of Munich, published Making the World Differentiable.

The paper proposed building a recurrent neural network (RNN), a neural network with two jobs: first, learn to predict what happens next in a simulated world and second, use that simulated world to train an Agent to act in it.

The Agent wouldn’t need to interact with a “real” environment at all. It could learn inside the model. Inside a dream.

The following year, Richard Sutton, of Bitter Lesson fame, dreamt up a similar idea. In Dyna, an Integrated Architecture for Learning, Planning, and Reacting, he argued that learning, planning, and reacting shouldn’t be separate systems. They should be unified in a single architecture. Which would mean that it’s technically possible to build a model of the world, practice inside it, and transfer what you learn back to reality.

Both papers were visionary. They would have a lasting impact as progress in the field enabled the researchers’ visions to become reality. But coming when they did, both papers may as well have been sci-fi.

In 1990, the world had something like 100 trillion to 1 quadrillion times less compute than we have today. Back then, the entire world had maybe 10-100 gigaFLOPS of total capacity. Tens of zettaflops (10^22 FLOPS) of computing power were sold in 2024 alone. In 1990, the global digital datasphere was approximately 10 petabytes, a volume so small it could barely hold 0.005% of the video data we now use for a single training run. By 2026, that volume has exploded by a factor of 22 million to 221 zettabytes.

But technology improves, and the most powerful dreams do not die.

Nearly three decades later, in March 2018, David Ha (then at Google Brain) and Schmidhuber published a paper titled World Models.4

The paper asked: Can agents learn inside of their own dreams?

To answer their own question, Ha and Schmidhuber built a fictional system with three components: a vision model (V) that compressed raw pixel observations into a compact representation, a memory model (M), a recurrent neural network that learned to predict what happens next, and a tiny controller (C) that decided what to do based only on V and M’s outputs.

The World Model was V + M: it could take in observations and imagine plausible futures. The controller was the Agent or policy: it chose which actions to take.

The paper joined in conversation with those centuries of thought experiments, novels, and movies. A dream might be reality, reality might be dreams. But what if we could actually act in our dreams? What would that do to reality?

Ha and Schmidhuber trained their World Model on observations from a car racing game and a first-person shooter game. The World Model generated new digital worlds. Then, they let the Agent practice entirely inside the World Model’s hallucinated dreams. Afterwards, they transferred the learned policy back to the actual environment.

And... it worked. The Agent could solve tasks it had never encountered in reality. The dream was real enough.

It was shocking, from a computer science perspective. But was it really so surprising? Isn’t this how humans navigate the world?

Ha and Schmidhuber noted that humans constantly run World Models in their heads. A baseball player facing a 100 mph fastball has to decide how to swing before the visual signal of the ball’s position even reaches their brain. The reason that every at-bat doesn’t result in a strikeout is that batters don’t react to reality, but to their brain’s “internal World Model’s” prediction of where the ball will be.

Donald Hoffman, Professor of Cognitive Sciences at University of California, Irvine, takes that idea a million steps further. He believes that we all walk around wearing “reality headsets” that simplify the staggering complexity of the quantum world into a user-friendly interface. Reality is too rich, so we navigate it via a sort of persistent waking dream.

This rabbit hole goes as deep as you want it to. But it’s World Models all the way down.

Ha and Schmidhuber showed that computers might be able to approach the world like we do: creating simulations to predict future states based on actions, acting based on those predictions, updating, and looping.

Actions, not words.

Language is Not Enough (Neither is Code)

Let’s play a game.

Clap your hands five times.

Now, instead of physically clapping your hands, I want you to describe clapping your hands using just words.

Where they are positioned in space, where they are relative to each other, by the picosecond. The points of contact. The sounds. What your hands look like as they move closer to each other, make contact, and pull apart. How they squish each other. What happens to the air between your two palms. What you see while your hands clap. Don’t forget your arms. How do they bend to facilitate the claps? Remember to do this by the picosecond, too. How does the fabric on your sleeve respond? What is happening in the background? Did the person next to you notice you clapping? How did they respond? Did you get fired for clapping in the middle of the meeting, following the instructions of an essay you shouldn’t have been reading while you should have been paying attention to work? Describe to me the vein on your boss’ forehead. Is it popping?

You can’t, can you? OK, stop. The point is made.

Language is an incredibly lossy compression of reality.

Language is important, of course. It is how we communicate and coordinate. The game Charades illustrates that to communicate ideas, language can be much more efficient than actions. LLMs are important in that capacity. But language alone is not enough.

What about code? Code is a form of very precise language that makes machines do things.

I asked Claude to “code me a simulation of hands clapping five times in a realistic environment.” It built me this. Which looks very painful.

There is a belief that, with scale, language and code will be able to solve all spatial-temporal intelligence challenges and produce Artificial General Intelligence (AGI) or Artificial Superintelligence (ASI).

Some argue that code is the key to solving many real-world intelligence challenges because it can perfectly instruct all physical form-factors with precision.

We do not share that belief. A code-based simulation is a poor version of a dream. It is rule-bound and unable to handle the stochastic messiness of reality.

To know the world, you must interact with it.

In The Glass Bead Game (Das Glasperlenspiel), a novel by Herman Hesse that won him the Nobel Prize for Literature in 1946, readers are introduced to Castalia, a future intellectual utopia devoted to pure thought. At Castalia’s center is an elaborate game, the titular Glass Bead Game, that synthesizes all human knowledge into a single formal language. Players compose “games” the way one might compose a fugue. A move might link a Bach cantata to a mathematical proof to a passage from Confucius. The game is the ultimate abstraction: all of human culture compressed into symbolic manipulation.

The protagonist, Joseph Knecht, rises to become Magister Ludi, Master of the Game, the highest position in Castalia. But he grows disillusioned. The game, for all its beauty, is sterile. Castalia’s intellectuals have retreated so far into abstraction that they’ve lost touch with the world. They can represent reality with extraordinary elegance, but they cannot act in it.

Knecht ultimately decides he must leave Castalia, and becomes a simple tutor. He chooses the messy, embodied, unpredictable world over the perfect symbolic one. He dedicated his life to the Game, the mastery of which involves operating on a level of abstraction beyond words, something closer to world modeling. But it wasn’t enough. Symbols alone, without contact with reality, eventually run dry.

Large Language Models are our Castalians. They are exquisite manipulators of symbols, capable of drawing connections across the entirety of human textual knowledge. They can discuss physics, compose poetry, write code, and explain the rules of baseball. They are, genuinely, one of the great intellectual achievements in human history.

But they operate entirely in the realm of representation. They can describe clapping, but they cannot clap. They can talk about gravity, but they do not know gravity the way a toddler knows gravity. They do not learn, the way a body learns, through thousands of falls and stumbles, what “down” means.

Language models predict the next token extraordinarily well. The only problem is that tokens are like shadows on Plato’s cave wall. And you cannot code your way to a realistic stadium crowd any more than you can describe your way there.

The real world is — or was — uncomputable.

If language and code, two of mankind’s most powerful inventions, are inadequate to represent our world, what do we have left?

The Answer is World Models

World Models offer another approach on the path to AGI. They offer a path to compute the things that are, today, uncomputable. They learn from the messy contact with reality that Knecht sought.

World Models offer a way to do non-deterministic compute efficiently, and to run simulations that shouldn’t be possible under traditional compute constraints.

World models are not a replacement for LLMs. Language remains essential; text can be used to condition World Models, to tell them what scenario to imagine, what goal to pursue, to give them a long-term goal. The thinking and the doing work together. But the doing has to come from somewhere other than text.

Joseph Knecht must come down from Castalia.

Real intelligence must come from observation of the world; from understanding actions and their consequences; from the things that language can only point at.

The Dao that can be told is not the eternal Dao.

In the beginning was the Word. Then came humans, to act imperfectly and unpredictably.

Maybe this is the way of things. In the beginning were LLMs. Then came World Models.

What Are World Models?

A World Model simulates environments and responds when you act inside them.

More formally, a World Model is an interactive predictive model that simulates spatial-temporal environments in response to actions.

While LLMs predict the next word in a sentence, World Models predict the next state (as in, the immediate future), conditioned on the current state and control input.

More succinctly: LLMs learn the structure of language. World Models learn the structure of causality.

This is a simple definition of World Models. It is accurate, but it’s not enough to understand how World Models work. For that, you’ll need to know four things:

1. What World Models do,

2. How they’re built,

3. Why “action” is so important, and

4. The relationship between World Models and policies.

What World Models Do

Think about what happens when you catch a ball. Your eyes take in a scene: the thrower’s arm, the ball in flight, the wind, the sun in your eyes, all of it. From that flood of sensory data, your brain builds a compressed model of what’s happening and, crucially, what’s about to happen. It predicts the ball’s trajectory a few hundred milliseconds into the future. Then it sends a motor command to your hand. You catch the ball. The whole loop — observe, predict, act — takes a fraction of a second and involves no language or “thinking” whatsoever.

A World Model does the same thing, computationally. It takes in observations (often video frames, though it can use any sensory data), builds a compressed internal representation of the environment’s state, and predicts how that state will change in response to actions.

It is, in essence, a learned physics engine, but one that doesn’t rely on hand-written equations. Instead of calculating gravity, collision, and friction from first principles, it has watched gravity, collision, and friction billions of times and learned the patterns.

This makes World Models a powerful tool for building Agents, AI systems that act in environments. World Models help Agents in three ways:

1. They serve as surrogate training grounds. An Agent can practice inside the World Model (basically, inside a dream) and transfer what it learns back to reality. This is important for safety (some things should not be tested or trained in the real world) and cost or sample/data efficiency (real world data is expensive, costly to gather, not available, you need a lot of it, etc.).

2. They enable planning over longer time horizons. An Agent can “imagine” the consequences of different actions before committing to one, the way a chess player thinks several moves ahead, except here, the board can be any environment or the real world.

3. They provide rich representations of the world for Agents to learn behaviors from. An Agent trained on a World Model’s internal representations learns to “see” the world in terms of the features that matter for acting in it, rather than raw pixels.

For these three reasons, the promise of World Models is that they are a path towards generalization. If you can create worlds that respond to actions the way the real world does, you can use them to safely, economically, and efficiently train embodied agents that can act in any virtual world, or the real one.

To be clear, this is the massive question in World Models: whether the simulated environments are faithful enough to reality that you can train on them and have that training transfer to the real world or more generally, whether you can “pre-train in sim.” Increasingly, the answer seems to be yes.

Ai2, the Allen Institute for AI, is a non-profit founded and funded by the late Microsoft co-founder, Paul Allen. It does great open source research and tooling, including its recent release of MolmoBot, an “open model suite for robotics, trained entirely in simulation.”

“Our results show that sim-to-real zero shot transfer for manipulation is possible,” they tweeted.

Dhruv Shah, a Princeton professor and Google DeepMind researcher who worked on the project, shared: “Within the scope of easily simulate-able tasks, a purely sim-trained policy outperforms SOTA VLAs trained on thousands of hours of real data!”

It is a pretty astonishing finding. A big focus of ours, and of the broader World Models field, is to expand the scope of tasks that are easy to simulate.

This is how it works. First, World Models imagine realistic environments and future states, ideally that respond to actions or instructions in the way the real and virtual worlds they’ve been trained on do. Next, the Agents are let loose inside of the generated worlds to train. Then, the Agents are brought back into real environments and are tested on what they’ve learned.

This is what Ha and Schmidhuber demonstrated in 2018. It remains the central promise of the field.

How World Models Are Built

World Models are fairly young. No single approach or combination thereof has proved superior, which means that the final architecture for general World Models is still an open question. There are, however, repeatable ingredients for training.

Start with data; massive quantities of observation data. Often, observations are paired with the actions taken to produce them. This pairing can come about in several ways. Observations (typically video) are collected in advance and actions are either recorded alongside them, or inferred via another model after the fact. Alternatively, the model learns by taking actions itself, generating its own observations and action data through direct interaction with an environment.

When the training data is observations or videos, the raw frames serve as observations of an environment unfolding over time. These videos are ideally labeled with the actions that produced them (either because they were recorded together or inferred with a separate AI model). The actions provide the causal link: what someone did that made the environment change. A gameplay clip where a player turns left and the camera pans to reveal a hallway. A driving recording where the wheel turns and the car follows a curve. A teleoperation session where a robotic arm reaches and a cup moves. In each case, the model sees a before, an action, and an after.

When the model learns through interaction, the same structure applies — before, action, after — but the data is generated on the fly rather than collected in advance, and the actions come from the model’s own developing policy rather than from an external source.

The World Model’s core objective remains the same: given the current state and an action or instruction, predict the next state. It sees frame t and action a, and tries to produce state frame t+1.

But predicting raw pixel worlds for everything can be expensive and often wasteful. Most of what’s in a video frame doesn’t change from one moment to the next; the walls stay where they are, the sky remains the sky. And most of the details within a frame are redundant; the color of the sky, the texture of a wall. They could be described in a more compact form.

So modern World Models involve a latent space: a compressed, learned representation where only the most essential information is retained.

The visual encoder compresses each frame down to a compact vector (a mathematical fingerprint of the scene) and the model learns to predict the next fingerprint — not every pixel in the 4K frame — in response to actions. This is where the computational efficiency comes from.

To accurately model the evolution of the world, World Models must also learn to represent the full set of possible outcomes. This uncertainty in outcomes is usually referred to as the stochasticity of the environment.

World Models have to learn to navigate what they don’t know yet (epistemic uncertainty: for example, a model that has never seen a traffic light will not know that red follows after yellow) and the inherently unknowable (aleatoric uncertainty: the randomness, like rolling dice5).

Even when the model has learned all that’s possible to know about the behavior of the environment (it has reduced its “epistemic” uncertainty to a minimum), there will almost always be some inherent uncertainty (“aleatoric” uncertainty) in what happens next. This is in contrast to pure entertainment video models, which only need to be able to predict a common evolution of the world state to perform well.

If you use a straightforward prediction approach (for example, a model naively trained with Mean Squared Error, or MSE) to predict a car turning a corner, the model can become ‘blurry’ because it averages every possible outcome. The car could turn and stay in the left lane, or it could merge into the right lane. The trajectory that actually minimizes the error is the implausible one where the car stays in the middle of the two lanes. That’s the blurriness, and different models handle it differently.

Diffusion models avoid this problem by gradually diffusing towards the outcome, enabling the model to commit to a specific mode of the outcome distribution, sampling a sharp, plausible future rather than averaging all possibilities.

Autoregressive models with multiple tokens per outcome also handle multimodality; by sampling one token after the other, they ensure that future token predictions are consistent with previous ones.

JEPA-style architectures, by contrast, address blurriness by simply sidestepping it. JEPA largely avoids having to model that distribution explicitly by never decoding back to pixel space at all. It operates in a space where averaging is less catastrophic, because we don’t expect these models to predict frames, but rather to develop representations that are useful for downstream tasks.

What comes out of this process depends on what you need. If you’re building a visual world simulator — something you can watch or explore — you decode the latent predictions back into pixels through a visual decoder, producing imagined video of plausible futures. This is what makes the demos from Google DeepMind and World Labs look realistic and impressive.

There are a number of approaches used to train World Models. We will cover them and how they evolved and built on each other through the lens of the brief eight-year modern history of the field shortly.

For now, keep this in mind: observation data in, paired with the actions that caused what’s happening in those observations, train World Models to predict the next state, Agents train to predict the next action in those Worlds.

Why Actions are the Ultimate Form of Compression

Here is a key insight behind World Models: actions are the ultimate form of compression.

Consider what happens when you decide to step left to avoid a puddle. Your brain processes the visual scene (the sidewalk, the puddle, the people around you, the curb, the approaching bus), predicts the immediate future (the puddle won’t move, the bus will pass, the person behind you will keep walking), evaluates options (step left, step right, jump, accept wet shoes), and selects one.

An outside observer can’t see inside your head, can’t know exactly what you were thinking, can’t know what you’re processing subconsciously. They don’t know if you’re tired or if you’re in a rush. They don’t know your moral code, how you, specifically, would answer the Trolley Problem. They don’t need to. They see the output of all of that near-instantaneous calculation: step left.

That, to me, is magic.

Of course, not everyone makes the right decisions. Play the video forward and you are able to learn the consequences, too. Step left, into an even bigger puddle. Step left, and get clipped by a car. Step left, and knock a baby out of its stroller. Over billions and billions of observations and instructions and actions, we learn not just how humans decide to respond based on inputs, but the consequences of those decisions. The collective World Model learns to act smarter than any individual.

Zoom back into the individual. If you could perfectly reconstruct someone’s stream of observations and actions, you would have a nearly complete record of their interaction with reality. You would know what they saw and what they did about it. The World Model learns exactly this mapping. It compresses space and time into a compact representation, and then uses actions to unroll what happens next. That’s what makes World Models so computationally efficient.

It’s also the same reason why World Models can handle stochasticity that traditional simulation cannot. To understand why, let’s revisit our Man U match with our new understanding of how World Models work.

In a traditional simulation engine, every possible behavior must be coded. If you want a thousand soccer fans to react realistically to a goal, you need to write rules for each type of reaction. The computational cost scales with the number of Agents and the complexity of their interactions.

In a World Model, the cost is fixed to one neural network pass. The stochastic, messy, human reality is already baked into the learned weights and absorbed from the millions of hours of video the model was trained on. The model doesn’t calculate what a crowd should do. It has seen what crowds actually do and it uses this information to make probable predictions.

This is what I mean when I call World Models compute for the uncomputable. Traditional computing is deterministic: known inputs, known rules, known outputs. The real world is not deterministic, so World Models don’t even try to code these things in. They watch, learn, and do, at a fixed computational cost, regardless of how complex the scenario gets.

World Models and Policies

There is one more distinction to make before we go further, one that gets muddled in typical conversations about World Models.

A World Model is a simulation of the environment; it takes in actions and produces predicted observations; it shows you what will happen if you do something.

A Policy is the brains of the Agent that acts within that environment. It takes in observations (and often instructions) and produces actions; it decides what to do.

The World Model is the dream. The Policy is the dreamer. The dreamer acts, and the dream responds. The dream responds, and the dreamer acts.

In practice, the relationship between the two turns out to be even more intimate and intertwined than that distinction suggests. Recent research has investigated training policies on top of World Model foundations or building them together from the get go. Start with the weights of a World Model — a system that has learned how to predict what happens next — and then, instead of training the model to predict future frames, or states, you train it to predict future actions.

A system that learns to predict the world can also learn much faster how to act in it. Understanding and doing aren’t two separate skills bolted together. They are the same skill, seen from different angles. At least this is what our research, and that of other labs, is starting to suggest.

That means that if you build a good enough World Model, you can also more effectively train a policy to act in the worlds it generates.

This is one of many important things the field has learned in a very short amount of time. Turns out intuition and imagination are two sides of the same coin.

A (Very Brief) History of World Models

On one hand, it should be very easy to summarize the modern history of World Models. It has only been eight years since Ha and Schmidhuber published World Models.

On the other hand, an awful lot has happened in just eight years. In that time, the field has gone through four waves: major periods when the field shifted its focus to prioritizing new questions. We highlight some of the most important papers here, and not boring world subscribers can find a full downloadable list of key papers at the end of the essay.

Wave 0, in 1990-1991, was the pre-deep learning era. Researchers first articulated the idea that Agents could learn internal models of the world and use them for prediction and planning. They asked, and answered, the question: what would a World Model do?

This is Richard Sutton and Dyna. This is Jürgen Schmidhuber and Making the World Differentiable. Before we had the compute, the data, or the architecture, we had the dream, waiting in dreamspace for reality to catch up.

Wave 1, in 2018-2019, asked: “Can this even work?”

Based on Ha and Schmidhuber’s work, the first paradigm involved using Video Auto-Encoders (VAE) to compress frames, model dynamics with Recurrent Neural Networks (RNN), and train policies inside the resulting dreams. So: compress what you see, predict what comes next, and train Agents to act inside that simulation.

At the time, the question was whether learning in imagination — dreams — was feasible. Researchers attempted to answer it using small models and simple environments to generate proof-of-concept results. Quite literally, the next big thing started out looking like a toy. Model Based Reinforcement Learning for Atari introduced the Atari 100k benchmark: whether the SimPLe algorithm could learn Atari games with only 100,000 real environment steps, or about two hours of gameplay.

The answer was yes. SimPLe learned how to play 26 Atari games and beat a competitor model on sample efficiency, or how many steps it took to reach a given score.

But could it play as well as humans?

That was the question that drove Wave 2 (2020-2022): “Can the World Model match human performance?”

DreamerV2, developed by Danijar Hafner at Google DeepMind, reached an answer quickly. They used a Recurrent State-Space Model (RSSM) with discrete latent representations — a system that maintains a compressed, running memory of the world and updates it with each observation. DreamerV2 became the first World Model Agent to achieve human-level performance across the 55-game Atari benchmark6. It was trained entirely in imagination, on a single GPU.

That same year, another DeepMind team published Mastering Atari, Go, chess and shogi by planning with a learned model in Nature. The paper described its MuZero model, which also beat Atari games (and others like Go), but did so by taking almost the exact opposite philosophical approach.

Whereas DreamerV2 generated observable dream environments and trained inside of them, MuZero never generated anything observable at all, planning entirely in abstract latent representations it invented for itself, and it did well.

It did so well, in fact that it leapfrogged the Go-specific models. In 2016, DeepMind’s AlphaGo beat human Go Champion Lee Sedol 4-1. It had been trained on a large database of human expert games plus self-play, with the rules of the game hard-coded in. The next year, AlphaGoZero beat AlphaGo 100-0 after being trained entirely from self-play with no human game data at all, just the rules. That same paper season, AlphaZero generalized AlphaGoZero’s approach to other games, like chess and shogi, both of which it came to dominate within hours. Then in 2019 (pre-print), MuZero learned everything, including the rules, the game dynamics, and the value function, from scratch, purely from observation and outcome. It matched AlphaZero on Go, chess, and shogi (where AlphaZero knew the rules) while also generalizing to 57 Atari games (where “rules” aren’t even a well-defined concept).

With each new model, something that humans had previously hard-coded — the rules, the strategy, the value of a position — was removed. The model learned each from scratch instead. MuZero was the terminus of that progression, entirely learned.

And MuZero did this without imagining future board states at all. It imagined hidden states, or abstract vectors it invented for itself during training that have no guaranteed correspondence to anything human-observable or interpretable. A human looking at MuZero’s internal representation of “three moves from now” would have absolutely no idea what it was thinking. And yet… it outperformed all previous models.

With MuZero’s success, the field now had two opposing schools of thought: generative World Models that produce observable futures, and latent World Models that predict in abstract space, even if they weren’t called “latent” yet.

From then on, progress in World Models has happened in both directions, generative and latent.

On the latent side, in 2022, Yann LeCun published a sweeping position paper from his dual positions at Meta and NYU Courant proposing a fundamentally different philosophy from generative models, one that looked more like MuZero: A Path Towards Autonomous Machine Intelligence. His new World Models company, AMI, is named after this paper.

LeCun’s Joint Embedding Predictive Architecture (JEPA) argued against generating pixels entirely. Similar to MuZero, instead of predicting what the world will look like, JEPA predicts what it will mean. It forecasts abstract representations of future states, deliberately discarding unpredictable visual details.

That same year, on the generative side, IRIS (2022), developed by Vincent Micheli and Eloi Alonso, two of General Intuition’s future co-founders, reframed World Modeling as language modeling over a learned vocabulary of image tokens. Instead of recurrent state-space models, IRIS used a GPT-style autoregressive transformer over discrete visual tokens. Basically, IRIS borrowed the machinery of language models and applied it to World Modeling.

In doing so, IRIS filled a number of previous gaps. The IRIS World Model was, in effect, a language model, but its vocabulary was images and actions instead of words. This brought the scaling properties of LLMs directly into World Modeling: efficient attention, scaling laws, and all the engineering infrastructure that had been built for large language models could now be applied to learning about the physical world.

Where Dreamer was missing the ability to model the joint law of the next latent state (for example, to handle multimodality), IRIS represented the next latent state as a series of discrete tokens to predict autoregressively, which meant that it was now able to predict multiple outcomes. And while Dreamer beat humans by using much more data than they do, IRIS was the first learning-in-imagination approach to beat humans with the same amount of available gameplay data (two hours).

JEPA aside, practically all of the work up to this point in World Models happened within games, and it’s worth floating between Wave 2 and Wave 3 for a second to appreciate the special relationship between AI and games.

Games have always played an important role in the development of AI. Claude Shannon’s 1950 paper Programming a Computer for Playing Chess is one of the founding documents of AI. In 1959, Arthur Samuel’s checkers program introduced the concept of machine learning itself. The first time the world woke up to the idea that intelligent machines could beat humans at anything was when IBM’s Deep Blue beat Garry Kasparov in chess.

Before DeepMind was an AI lab, Demis Hassabis was a game designer. At 17, he designed the commercially successful Theme Park. DeepMind’s founding breakthrough is detailed in the DQN paper, published in Nature in 2015, in which it was demonstrated that Atari games could be played from raw pixels using deep reinforcement learning. Then came AlphaGo in 2016, which beat the world champion at Go, a game that once was believed to require the kind of intuition that was uniquely human, with more possible board positions than atoms in the universe.

The path from AlphaGo to AlphaFold ran through exactly the insight that World Models formalize. As Hassabis put it:

Wouldn’t it be incredible if we could mimic the intuition of these gamers, who are, by the way, only amateur biologists?

General Intuition is named after this quote from Demis, which points towards a future where our models power research far beyond the dynamics of what pixels can describe today, beyond games themselves, and into our bodies.

And then DeepMind taught machines how to fold proteins. AlphaFold won Hassabis and his DeepMind teammate John Jumper the 2024 Nobel Prize in Chemistry.

Games are fun, of course. But the reason games keep showing up is that games are the only domain where you get massive amounts of labeled spatial-temporal data with clear action-outcome pairs, consistent physics, unambiguous reward signals, and a controlled environment where you can run millions of experiments. The real world has none of these properties.

Early World Models, like a human child, spent most of their time watching and playing games. The Atari 100k benchmark became the standard arena for World Model research, DreamerV3 played Minecraft, and many current World Model companies retain a connection to games, with many World Models being “playable.”

Games are the lab bench of embodied AI. But they are only a small fraction of the ambition.

For World Models to be truly useful, they need to interact with the world.

That’s Wave 3 (2023-2024). It asked: “Can World Models be truly interactive?”

We got the first answer from driving. GAIA-1 (2023), developed at Wayve, scaled the sequence-modeling approach pioneered by IRIS to 9 billion parameters and trained on real-world driving video. It could generate driving scenarios in response to actions (steer the car), text prompts (“rainy day, highway”), or both. Anthony Hu, who led this research, now leads World Modeling at General Intuition.

GAIA-1 confirmed that the scaling laws everyone had observed in LLMs also held for visual World Models. More data and more parameters yield predictably better performance for World Models, too. This was not a given. It meant that the path forward was clear even if it was expensive: scale up and the models get better.

The following year, DIAMOND (2024), developed by future General Intuition co-founders Eloi Alonso, Adam Jelley, and Vincent Micheli, opened a new architectural frontier. Rather than compressing observations into discrete tokens and predicting them autoregressively, as researchers had been doing since IRIS, DIAMOND used diffusion models to predict future frames directly.

The visual fidelity was meaningfully richer, and that richness translated directly into better Agent performance. The subtle visual details that discrete tokens discarded, the little clues that tell you a surface is slippery, a door is ajar, a person is about to change direction, turned out to matter for decision-making, which is unsurprising when you think about it.

As a brief aside, it’s worth noting that many of the open source advancements that have been made in World Modeling were built on top of the DIAMOND architecture. Multiverse, the first AI-generated multiplayer game, is DIAMOND-based, as is Alakazam, the “1st ‘World Model game engine’.” DIAMOND is essentially the Deepseek or Llama of Generative World Models.

DIAMOND itself set a new best on Atari 100k and demonstrated something that captured the public imagination: trained on Counter-Strike gameplay, it produced a fully interactive, playable neural game engine from roughly 87 hours of footage on a single GPU.

It showed that it was possible to run an interactive 3D World Model in real time too.

Counter-Strike environment generated by DIAMOND with just 87 hours of footage

DIAMOND got very good at playing Atari. The Agent plays a real game and gathers real data there, with which it trains the World Model. Then it tests itself inside the World Model’s synthetic environment, gets better in there, and goes back out for more real interaction, to test itself in the wild. This loop between ground truth and synthetic, back and forth, is how World Models improve, almost like working problems out in a lucid dream then testing them in reality upon waking. This is the Dyna paradigm mentioned earlier.

Would that loop work in real-world conditions?

It turns out that the answer is yes, too. And that it would work beautifully.

GAIA-2 (March 2025) pushed the diffusion approach to its most ambitious application yet: multi-camera autonomous driving simulation. Using latent diffusion with flow matching and space-time factorized transformers, the model could generate high-resolution surround-view driving video conditioned on ego-vehicle dynamics, other Agents’ trajectories, weather, time of day, road structure. In short, it could reproduce the full complexity of real driving. It could simulate scenarios that were too dangerous or too rare to collect from real roads: sudden cut-ins, emergency braking, pedestrians stepping off curbs.

GAIAs 1 and 2, and DIAMOND, like IRIS, were the products of researchers we now get to work with at General Intuition. Diffusion or flow-matching models like GAIA-2 were the starting point of our team’s current research efforts.

But they are not the only approach.

Google DeepMind is one of the central players in this space. Their World Model, Genie (2024), is an 11-billion-parameter model trained on unlabeled internet video of 2D platformer games. It learned an action space entirely from scratch; no one ever told the model what the controls were. Give it any image and it can generate a playable world from it.

OpenAI’s Sora (2024, with Sora 2 following in 2025) and Google’s Veo 3 (2025) pushed video generation to extraordinary visual quality and framed these systems explicitly as “world simulators.”

The field’s vocabulary can get muddied. Let’s make it clear.

Video generation models produce beautiful visual sequences, but they aren’t quite World Models in the sense that we’ve been describing them. In these videos, you can’t take an action and watch the environment respond live to your intervention. They predict what a scene will look like over time; they don’t model what happens because of what you do.

Think of the difference between watching a movie of someone driving and actually steering a car. The visual output might look similar, but the underlying computation is fundamentally different. Interactivity, the ability to take actions and observe their consequences, is what separates a World Model from a very impressive video.

And interactivity is what it takes to impact the real world.

This is the central question of Wave 4, the wave we’re in right now: “Can models act in the real world?”

As in: Can Agents trained in World Models work outside of research settings, in real vehicles, real robots, real deployments? We are now getting awfully close to sci-fi’s predictions.

This is where the current frontier is being pushed. Right now. As you read this.

Comma.ai took the most direct path in driving from World Model to product: Learning to Drive from a World Model. They trained a driving policy entirely inside a learned World Model — inside the dream — and deployed it in openpilot, their open-source driver assistance system running on production vehicles driven by real people. The World-Model-trained policy outperformed both traditional imitation learning and policies trained in conventional simulators. This is arguably the first consumer product powered by a World-Model-trained Agent.

In robotics, Meta’s V-JEPA 2 animated LeCun’s latent prediction philosophy. The model is the clearest large-scale proof point so far. It’s a 1.2B-parameter model pre-trained on over a million hours of video via self-supervised masked prediction: no labels, no text. In the second stage, it fine-tunes on just 62 hours of robot data from the Droid dataset. It turns out that is enough to produce an action-conditioned World Model that supports zero-shot planning. V-JEPA 2 was deployed zero-shot on real Franka robot arms in new environments to perform pick-and-place tasks. It planned all of this entirely in latent space, without pixel generation, task-specific training, or hand-crafted rewards. And it was fast; where pixel-space approaches took minutes to plan a single action, V-JEPA 2 did it in seconds.

Google DeepMind’s SIMA 2 took an entirely different approach. Rather than build a dedicated World Model, it fine-tuned Gemini, its large foundation model, to act directly as an Agent in 3D game environments. SIMA 2 can reason about high-level goals, follow complex multi-step instructions, converse with users, and generalize to unseen environments.

It represents an alternative paradigm: instead of building a specialized World Model, leverage the implicit world knowledge already embedded in a model trained on the breadth of human knowledge.

This is one of the field’s open questions. Will this path, using a large foundation model or a video model as the basis for an Agent, rather than training an Agent from scratch in a World Model, win out?

In fact, there are many open questions. And nearly as many World Model startups trying to answer them.

The State of the World (Models)

That brings us to the present moment.

What has become clear is that talented researchers and investors alike are excited by World Models’ potential, as evidenced by the massive funding rounds to support companies led by legends in the field.

In February 2026, World Labs, the company founded by legendary researcher Fei-Fei Li, announced that it had raised a fresh $1 billion from investors at a $5.4 billion post-money valuation.

Not to be outdone, Yann LeCun, who launched AMI Labs in late 2025, announced last week that it had raised $1.03 billion at a $3.5 billion valuation.

In October 2025, our company, General Intuition, announced $133.7 million in a very large seed round. Last summer, Decart raised $100 million at a $3.1 billion valuation. In November, Physical Intelligence raised $600 million at a $5.6 billion valuation for its robot foundation models. And just this past February, Wayve, the UK-based self-driving startup whose researchers built GAIA-1 and GAIA-2, raised $1.2 billion at an $8.6 billion valuation.

Google DeepMind, which doesn’t need to fundraise because it’s fueled by history’s greatest business machine, is pouring resources into SIMA, Genie, and Veo, and using it to power initiatives like Waymo. Demis has publicly stated that he believes World Models will become an important part of Gemini’s planning capabilities. GDM is also merging many of these capabilities into a “Video Thinking” team, with the reasoning described best by Shane Gu and Jack Parker Holder from GDM.

What is less clear, but even more interesting, is that we are at the point in this technology’s development where we know that something big is happening, but it’s still unclear exactly which approach, or combination of approaches, will win out. We are seeing breakthroughs almost every day at General Intuition, and we hear rumors of leaps happening in other labs too.

Below is a framework in which to fit any news you see coming out on World Models. We won’t cover everything, and we apologize in advance if we miss your embodied AI of choice. A fun exercise for the reader will be to fit what we’ve missed into what we’ve laid out.

World models have three main types of approaches: Current Foundation Models, World Models, and Embodied Agents.

The thing to keep in mind here is that, despite different World Model approaches, we all share the same end goal. The end goal is to produce Agents that generalize and do things in various environments, including the real world. Some of the Agent approaches get there using LLMs as their stepping stone, others start with video models. Other agent approaches use World Models as their training environments. And some Agents learn directly from experience.

With us? Goed, daar gaan we dan!

Current Foundation Models

Current Foundation Models are the ones that learned to make sense of the world’s data without being able to simulate the stochastic world environment itself. They are models that process inputs — text, images, video — and learn to predict, generate, or reconstruct. But they don’t yet give an Agent a place to act. They are not action-conditioned. They don’t respond or interact. They are potential substrates on which World Models can be built, or even, in some cases, on which Agents are pre-trained.

Three categories of stepping stone models we’ll focus on here are Large Language Models, Video Models, and 3D Reconstruction Models.

Large Language Models

LLMs learned from staggering quantities of text that the world has structure. They know that a glass falls when pushed, that fire is hot, that if you leave the house without an umbrella in a rainstorm you will get wet. They encoded an enormous amount of causal and physical knowledge. But none of this was from experience. Like digital Castelians, they read about the world rather than perceiving it. This makes them extraordinarily useful as a backbone for reasoning and planning, which is why you’ll find LLMs embedded in many agent architectures we’ll discuss later. But a language model alone cannot simulate what happens when a robot arm reaches for a cup.

In our context, LLMs are particularly relevant when we discuss VLAs, or Video Language Action models, which take advantage of the enormous amount of research, capital, tooling, and infrastructure that has gone into developing LLMs in order to bootstrap robots that can do things in the physical world.

Video Models

Sora. Veo 3. Kling. Seedance 2.0. Runway. Pika. Moonvalley. Haiper. Luma AI.

No one confuses an LLM for a World Model, but plenty of people conflate Video Models and World Models.

These models are trained on the enormous amount of video data on the internet, and produce extraordinary videos themselves. Sora can generate a convincing shot of a woman walking through a neon-lit Tokyo street. Veo 3 can render photorealistic scenes with synchronized dialogue.

But you can’t interact with them. You can’t take an action inside of them and watch the environment respond instantly. They predict what a scene will look like over time but they don’t try to model what happens because of what you do.

Of course, the lines get blurry.

Odyssey, founded by self-driving heavyweights Oliver Cameron (ex-Cruise) and Jeff Hawke (ex-Wayve), is building “a world simulator that dreams in video.” Currently, they don’t let you take an action and watch the environment respond, but they do let you prompt the video mid-stream to steer it in real-time. Where do you draw the line?

Wherever the line is, these video models are getting good, and really funny.

Really, really funny.

Video models aren’t quite World Models in the sense we define them; they are a stepping stone. Runway began as a video generation company – its Gen 4.5 is among the best on the market – but has concluded that physics-aware video generation is a path toward something bigger. This thinking led to GWM-1, their explicitly labeled “General World Model, built to simulate reality in real time,” which is interactive, controllable, and general-purpose. The real value, financially and societally, won’t come from video for its own sake, but from models that use video as a training environment on the way to controlling embodied systems.

3D Reconstruction and Generation Models

Take it a step further. What if you could navigate through the scenes depicted in video generation models? That feels like a world, right?

World Labs, led by the legendary Fei-Fei Li, the “Godmother of AI” who created ImageNet, is the most interesting example in this category. While the company is the one most people would associate with “World Models,” World Labs is not currently building what I would define as World Models.

Instead, in its early days, World Labs has focused on immersive virtual worlds, but not action-conditioned ones. Its first product Marble generates and edits persistent 3D environments from text, images, video, or 3D layouts. They call it a “Multimodal World Model.”

Marble is thus far not interactive, other than being able to move through the generated environments. They say this themselves. On the Marble product page, World Labs frames interactivity as a future opportunity:

Future World Models will let humans and agents alike interact with generated worlds in new ways, unlocking even more use cases in simulation, robotics, and beyond.

It is worth noting that World Labs has recently started exploring World Models that do generate frames directly, instead of the underlying splats of the entire world.

World Models

A World Model, as we define it, is an environment that an Agent can act in, and that responds in real time. It is a simulation, a dream, one learned from observation and actions data rather than hand-coded. The Agent takes an action, the world changes, and the Agent observes what happened. Repeat, millions of times, across an enormous variety of situations, and the hope is that you get an Agent that generalizes, that can do things that were not in the original training data.

This is the key distinction that everything else hinges on: a World Model is action-conditioned. It predicts what the world will look like next given whatever the Agent did.

The intuition is simple. A robot trained only on real-world data has seen a finite set of kitchens, a finite set of cups, a finite set of ways a cup can fall. Put it in a kitchen it hasn’t seen, with a cup it hasn’t encountered, and it struggles. A robot trained inside a World Model, on the other hand, has, in principle, encountered infinite kitchens because the World Model can generate them. Situations that would be rare, expensive, or dangerous to collect in the real world become routine in simulation. Out-of-distribution becomes in-distribution.

Within World Models, there are two main approaches: Latent World Models and Generative World Models.

I apologize for bringing you so far in the weeds here, but I want to clarify something that confuses people: both Generative World Models and Latent World Models rely on latent states, but Generative World Models rely on latent states that were designed with reconstruction objectives (autoencoders) which enable frame predictions, whereas Latent World Models directly build self-predictive representations.

Latent World Models were born in the darkness and still live there; Generative World Models were merely born in the darkness.

Latent World Models

Latent World Models are the descendants of MuZero but let loose in open-ended, no-rules environments like the real world.

This is Yann LeCun’s current world. Yann pioneered modern computer vision architectures with LeNet, where he introduced the idea behind convolutional neural nets (CNNs) in the 1990s. In the 2010s, he championed self-supervised learning, arguing that human labeling millions of examples doesn’t scale to real intelligence and that models should create their own signal from raw data. In the 2020s, he led the JEPA team. Yann is a GOAT.

The deep thread in Yann’s work is teaching models to learn useful representations of the world automatically from raw data. Latent World Models are the latest, and perhaps ultimate, strand in this thread.

The approach is philosophically the converse of Video Models or 3D Reconstruction Models, as mentioned earlier in the history section. While those approaches care about producing and understanding every pixel, latent World Models, like JEPA, says ne vous embêtez pas. The French would rather speak English to you than listen to you butcher their language. JEPA is similarly impatient; rather than let the model stumble over every pixel of an unpredictable future, it doesn’t predict pixels at all.

As LeCun puts it: “The world is unpredictable. If you try to build a generative model that predicts every detail of the future, it will fail. JEPA is not generative AI.”

Instead, JEPA learns to represent videos in abstract, compressed space and makes predictions there. It deliberately throws away unpredictable visual details. This makes JEPA potentially very efficient for planning and representation learning.

AMI Labs is LeCun’s bet that this approach is the path to real intelligence, and investors recently backed him with $1.03 billion.

There are trade-offs to the latent approach, as there are trade-offs to generative approaches.

LeCun argues that the thing that seems like the biggest trade-off, a loss of fidelity in exchange for speed, is not actually a trade-off. His position is that the detail you lose is detail you should lose, that trying to predict every pixel is not just expensive but actively counterproductive — the model wastes capacity on inherently unpredictable visual details instead of learning the abstract causal structure that actually matters for reasoning and planning. Imagine if you needed to simulate every photon when you imagine catching a ball. Your brain might explode. There’s some level of detail that is not “every single detail” that is optimal. LeCun’s argument is that with World Models, the optimal level requires fewer details than many people, including us, think.

There are other trade-offs to keep in mind, however, that LeCun hasn’t mentioned.

One is that latent models are trickier to evaluate. You can’t look at the output and see if it makes sense intuitively the way you can with generated video, and they can’t serve as training grounds for human-in-the-loop systems, because humans can’t operate inside latent space. We need to see the world to act in it.

Another related downside is that your iteration speed slows down when you can’t visualize predictions or interpret the loss. Humans are very good at noticing when something is visually off; we did not evolve to spot discrepancies in predicted latent encodings of the future ([0.13, -1.02, 0.44, 0.07, …], MSE = 0.0187). And iteration speed is what matters most in modern ML, because modern ML progress mostly comes from empirical search, not from knowing the right design ahead of time.

Latent models are also more challenging to train for similar reasons. Additionally, the lack of strong supervision in the learning objective leads to collapse issues which require a bunch of tricks to fix. Why? The JEPA objective is to predict the encoding of the future based on the encoding of the past, but you can satisfy that objective with trivial encodings (e.g. set everything to 0, there is 0 loss), so we need to make sure representations don’t collapse.

There is a spectrum in creating environments in which agents can train. On one side is what is practical today, and on the other is the platonic ideal. Latent World Models are almost the opposite side of the Practical ←→ Platonic spectrum to VLAs, which we will cover below.

They are closer to what researchers believe to be the technical platonic ideal, but they face real challenges in practice today. That said, new methods like LeJepa are closing the gap, and talent is flooding into the field.

Chris Manning, Ian Goodfellow, and Fan-Yun Sun have also joined the cohort of Latent World Models, starting latent lab Moonlake. Their entrance on the side of latent is notable. Manning helped pioneer neural natural language processing and co-created GloVe, which was the dominant word-embedding model before transformers. And Goodfellow invented GANs (generative adversarial networks), which were the first widely successful way to train neural networks to generate realistic synthetic data.

In a recent X post, the Moonlake co-founders explained their approach to building efficient World Models. It is an interesting hybrid.

The plan is to generate full game environments to attract human players and collect action-labeled data. Afterwards, they model the world in semantic/symbolic space rather than pixels. As in, they use beautiful game environments to attract real human players, because they need humans to generate action-labeled data. But once they have that data, they discard the pixels entirely and train on abstract representations instead, betting that the underlying patterns matter more than the visual detail.

Ultimately, we don’t view latent and general models as opposed to each other. Moonlake’s hybrid approach is evidence of that. They just serve different goals. Latent World Models are generally more computationally efficient since they discard some information, with advantages for representation learning and planning. Generative World Models should be more general, since in theory they capture all visual information, with advantages for interpretability and generalization. Both can be used for many different purposes, including training agents with reinforcement learning.

Now, let’s turn to Generative World Models.

Generative World Models

Generative World Models are the closest thing to simulating human-perceived reality that we’re aware of. If our world is a simulation, it’s probably a Generative World Model of some sort.

This is the paradigm we at General Intuition predominantly focus on to build the World Models in which our policies learn. It’s also the one that recently blew the world’s minds when Google DeepMind released Genie 3.

The video — and, Genie 3 itself, if you get a chance to play with it — gives you a felt sense for what makes Generative World Models different. They’re interactive. They respond.

They generate human-observable, interactive futures that you can see, act in, and learn from. You can see what the model thinks will happen next. The model takes in a state and an action and produces a plausible next state, which you can act in again. Based on the updated state and new action, it produces the next plausible next state, and so on and so forth. A human can look at the output and say, “That’s wrong, walls don’t bend like that” or “Yes, that’s exactly what happens when you turn a steering wheel at speed.”

Generative World Models predict the observations themselves in pixels, video, or 3D scenes, allowing agents and humans to interact with the simulated environment. The dream is visible and playable.

This is what improves the training loop in many cases. Both generative and latent models can learn in imagination. Yet when visual details matter, or when the downstream task is not yet known, Generative World Model learning, with all its pixel-level detail, tends to outperform.

This only works if the generated environment is rich enough to learn from. The further the generated world is from reality, the worse the lessons the agent learns are. And the less successful it is when it goes back to real games. This is what DIAMOND showed, that when there is more detail in the generated world, agents are smarter.

At General Intuition, we are building on this diffusion and flow-matching architecture. It is developed, in part, by researchers who are now our co-founders and who built IRIS, DIAMOND, and GAIA-2.

Wayve, the birthplace of GAIA-1 and GAIA-2, is the leader in Generative World Models for autonomous driving. By using a large latent diffusion World Model offboard, they aim to dream up edge cases that would take millions of driving miles to find in reality, train driving policies on them, score the driving policies on their performance in simulation, then distill that dreamed experience into a smaller onboard policy that can reason through those same scenarios in real-time. The tweet below shows Wayve zero-shotting a drive on Japanese roads in the latest installment in a series doing this around the world.

Decart is applying Generative World Models to real-time generative simulation, producing playable worlds that respond to the users’ actions. It’s the playable version of the Generative Video Models or 3D Reconstruction Models. On the Oasis landing page, it calls the model a “video model,” but follows up with this distinction: “Every step you take will reshape the environment around you in real-time.”

Interestingly, Decart currently runs on Nvidia GPUs but plans to use Etched Sohu chips. Etched chips are custom ASICs designed to run transformers and would allow Decart to improve latency and run continuous inference, both of which are much more important when generating responsive worlds in real-time than when generating a video or 3D rendering upfront.

Runway, too, is blurring the lines between video generation and world generation, as mentioned in the Video Model section. During its Research Demo Day 2025, Runway co-founder and CTO Anastasis Germanidis explained the company’s evolution that started from “generative AI models [as] viable tools for creative expression”. They then evolved towards World Models (while still making incredible progress in video models.)

“To build a World Model,” Germanidis explained, “We first needed to build a really great video model. We believe that’s the right path to building World Models, that teaching models to predict pixels directly is the best way to achieve general purpose simulation.”

Google DeepMind took a similar approach; Genie 3 was built on top of Veo.

These World Models are extremely important. But remember that they are only half of the equation. From the very start, whether it was Schmidhuber in 1990, Sutton with Dyna in 1991, the plan was to use World Models to train Agents to act inside of the world, and then to transfer those learnings out into the real world.

Embodied Agents

We want to share a few of the main Embodied Agent examples out there today, and their respective approaches: Physical Intelligence and other robotics companies’ VLAs (Vision-Language-Action Models), DreamerV4’s a Latent World Model Agent, Google Deepmind’s Sima2 General Embodied Agent, and General Intuition’s General Agent approach.

Physical Intelligence - Vision-Language-Action Models (VLAs)

Modern multimodal LLMs come with a model called a VLM, or Vision-Language Model, a model that can see and read. Feed it an image and a question, like “What objects are on the table?” or “Is this door open or closed?”, and it produces a coherent, grounded answer.

GPT-5, Gemini, and Claude are all VLMs in this sense; they can see and reason. When you send it a picture of a mountain and ask it to geolocate it, it’s using its VLM. VLMs are also the perceptual and reasoning backbone of most modern Agent systems designed to operate in physical or interactive environments, like PaLM-E or SpatialVLM.

VLMs are not exactly Agents, but they are core components of most of them. We bring them up because a VLA is a VLM that has learned to act, and it is the pragmatist’s answer to the Agent problem.

In 2023, Google DeepMind published a paper called RT-2: Vision-Language-Action Models: Transfer Web Knowledge to Robotic Control to propose a solution.

Take a VLM that understands a scene and what to do in it, and then bolt on an action head that translates human language instructions into instructions that the robot understands, like to change a position or rotate.

Since then, VLAs have become the dominant paradigm in robotics, and they’ve worked surprisingly well.

Every other paradigm we’re discussing says something like: “Images, videos, spaces, and actions are fundamentally different than words. We need to train and architect the models that generate them differently than the models that generate words.”

Vision-Language-Action Models (VLAs) say: “That may be true! Those approaches might be platonically better. But that doesn’t matter in practice, because the vision-language model infrastructure and data are so far ahead.”

In his not boring primer on robots, Standard Bots’ Evan Beard wrote a thorough explanation of VLAs for robotics that included what he called a “spicy take”:

We’re not working with language-model infrastructure because it’s the perfect architecture for robotics. It’s because we, as a species, have poured trillions of dollars and countless engineering hours into building LLM infrastructure. It’s incredibly tempting to reuse that machine.

So, despite its imperfections, taking an LLM and sticking on an action head to predict robot motions (all together known as a VLA) is the best way for us to train the base models that learn many skills from demonstrations across many different customers and tasks.

It’s pretty ingenious. Of course, there are challenges to this approach, which Evan highlighted:

• Robotics success so far has leaned heavily on diffusion-style control

• LLMs are autoregressive and token-based, with less room for error

• Physical actions don’t map cleanly to tokens

Additionally, compared to World Models, VLAs require collecting a large amount of real-world robotics data; they don’t seem to generalize out-of-distribution particularly well.

That said, Physical Intelligence, known as π or Pi, has gotten incredibly far with its VLA bet.

Pi’s first generalist policy, π₀: Our First Generalist Policy, inherits semantic knowledge and visual understanding from internet-scale pretraining and trains on data from seven different robotic platforms across 68 unique tasks, including folding laundry, bussing dishes, routing cables, assembling boxes, and packing groceries, all of which require dexterity in the real world on real hardware. Their follow-up, π₀.5: a VLA with Open-World Generalization, performs better in new environments like cleaning up a kitchen or bedroom in a home the model has never seen before.

OK, but can it actually learn and get better over time as it works and makes mistakes in the real world?

November 2025’s π*0.6: a VLA that Learns from Experience suggests that it’s possible, with demonstrations in tasks like making espresso, folding boxes, and folding laundry.

But those are simple, repetitive tasks. Most of what the robot sees is in distribution. Can it actually do more complex, multi-step tasks that take a long time to complete?

Earlier this month, Pi released VLAs with Long and Short-Term Memory and showed that robots using MEM (Multi-scale Embodied Memory) can clean up an entire kitchen, set up the ingredients for a recipe, and grill a grilled cheese sandwich. They can also learn from their mistakes.

A robot tries to pick up a chopstick or open a refrigerator door. Without memory, it fails the same way repeatedly. Each attempt is a clean slate with no knowledge of what just went wrong. With memory, it tries a different approach after the first failure. And it succeeds.

MEM doesn’t change the underlying architecture, which is still sub-optimal for embodied systems. Most of the parameters still live in the language backbone. The action head is still downstream of reasoning. But Physical Intelligence’s existence raises a fascinating question. Do these architectural limitations actually matter in practice?

If Latent World Models are on one side of the Platonic ←→ Pragmatic Spectrum, VLAs are on the other.

To date, Pi has been able to engineer their way around architectural limitations to make increasingly capable robots. Their progress is not slowing down. It seems to be accelerating.

Theirs is a bet with historical precedent. The ideal technology — the solution that is technically superior — does not always win. This is the key takeaway from W. Brian Arthur’s 1989 paper, Competing Technologies, Increasing Returns, and Lock-In by Historical Events. Markets often converge on the technology that gets adopted first, because adoption creates increasing returns: better early product means more users and more capital, which means better data, more internal talent, and more developers, which means better products which means more users and capital, and so on.

This is also the point of Sara Hooker’s 2020 paper, The Hardware Lottery: “This essay introduces the term hardware lottery to describe when a research idea wins because it is suited to the available software and hardware and not because the idea is superior to alternative research directions.”

From the outside, it seems that Pi’s strategy is to ride the transformer architecture’s increasing returns and trying to generate its own, to create path dependence with VLAs before World Model-specific architectures gain traction, in an attempt to win its own hardware lottery.

They’re not the only company making this bet. Skild, the closest direct competitor, is building on VLAs. A number of robotics companies incorporate VLAs and VLMs in one way or another. And now, it looks as if the approach is spreading to the whole factory.

Recently, the WSJ reported that former OpenAI Chief Research Officer Bob McGrew is raising $70 million at a $700 million valuation for his new startup, Arda, in a round led by Founders Fund and Accel, with participation from Khosla and XYZ. Details are light, but the WSJ’s description sounds like it’s going to at least involve VLMs and VLAs in some way: “Arda is developing an AI and software platform, including a video model that can analyze footage from factory floors and use it to train robots to run factories autonomously.”

The more well-funded and talented companies that move in this direction, the more deeply grooved the path becomes.

Personally, I don’t think VLAs and World Models are really competing. They’re trying to reach acting in the physical world from different directions. VLAs are language first, while World Models are video actions-first. My guess is that they’ll converge and both be part of the solution.

Dreamer V4 - Latent World Model Agents

Latent World Model Agents are Agents trained inside of Latent World Models. The latent approach has a natural elegance for Agent training specifically.

Because a Latent World Model operates in compressed abstract space, the Agent’s planning and policy learning can happen very efficiently, with no pixel generation required. The agent basically practices by thinking, in the same way a chess grandmaster runs through variations in their head without moving the pieces, or the way a lucid dreamer trains inside the dream.

The canonical example is Dreamer, from Danijar Hafner, now at Google DeepMind. Dreamer’s insight is elegant: if you have a good enough Latent World Model, you don’t need to touch the real environment during training at all. The Agent imagines sequences of actions and their consequences entirely in latent space, receives a reward signal, and updates its policy, all without a single real-world interaction. When it finally goes into the real environment, it already knows what to do.

Dreamer has achieved remarkable results across a wide range of tasks, from games to continuous control to robotics, all from this purely imagined training. It is the research proof of concept that World Model training works, that an Agent can learn to act in the real world by dreaming. It seems as if Hafner is taking his research proof commercial. Earlier this month, The Information reported that he and Wilson Yan are raising $100 million to build a World Model company in this paradigm called Embo, which suggests they’re going after embodied systems.

The challenge, as with Latent World Models generally, is that the Agent’s learned behavior is only as good as the latent representation. If the World Model’s abstract encoding misses something causally important, like the exact texture of the floor that determines whether the robot slips or the precise angle of an object that determines whether it can be grasped, the Agent won’t know to care about it, because the model didn’t encode it. Garbage in, garbage out, but the garbage is invisible.

Moonlake’s hybrid approach, which we discussed earlier, is an attempt to thread this needle: attract humans with beautiful generative environments to collect action-labeled data, then discard the pixels and train the Agent in abstract space. Use the generative world to get the data. Use the latent world to do the learning. It’s an interesting bet that the two approaches are more complementary than competing, and it may prove correct.

Notably, we haven’t yet seen the JEPA Agents. JEPA is a World Model architecture, not an Agent architecture, but we expect AMI Labs will close this loop. AMI is still building its World Model, and the Agents that train inside it haven’t yet been publicly demonstrated, but we’re watching closely.

General Embodied Agents

SIMA2 - Generalist Embodied Agents from VLM Backbone

In November 2025, Google DeepMind released SIMA 2: An Agent That Plays, Reasons, and Learns with You in Virtual Worlds.

SIMA 2 combines a Gemini backbone with a World Model trained on 3D game environments, giving the Agent an understanding of language that allows it to receive and reason about goals, as well as the spatial-temporal understanding to execute on them. In this architecture, Gemini fills the role that we mentioned VLMs play in our system.

What makes this a different paradigm from VLAs is the direction of citizenship. In a VLA, language is first class, images are second class. Beyond the ordering of modalities, there is also the training data, mostly static images interleaved with text. In an Agent equipped with a World Model, video is first class, actions are introduced from the beginning, and the training data is directly aligned with the downstream behaviors we’re looking for. The Agent’s fundamental competence is spatial-temporal. If you tell it what it needs to do, it knows how to move through the world to do it.

SIMA 2 can play games on its own. It can learn, reason, and improve. The more it plays, the better it becomes, not just in the games it’s played, but in any game it plays. It is even able to play in whichever generated world it gets thrown into, even when it’s never seen it before. This, Google DeepMind believes, is a “step towards creating AI that can help with any task anywhere, including one day in the real world.”

Google DeepMind has put out an extensive amount of research. They have pushed World Models and embodied AI forward on multiple fronts. They coined the term “VLA.” They released Genie 3. They developed SIMA 2. The way they trained AlphaGo, letting the Agent play against itself over and over and over again, informed how World Models are trained to this day.

General Intuition - Generalist Agents From Actions & World Models

Similarly to Google DeepMind, we also believe that Generalist Agents will play a major role in how embodied systems operate to do useful things.

First, create the dream. Then, let Agents run around inside of it. Let them play and mess up and learn and win. Then, transfer those learnings into other dreams, and even into the real world.

Recall the Matrix. When Neo needed to learn Kung Fu, he plugged into a virtual dojo where he trained against Morpheus in a training environment superior to the “real world.” After that? “I know Kung Fu.” World Models are the virtual dojo. Neo is the Agent.

This is the question that Ha and Schmidhuber asked eight years ago: can Agents learn inside of their own dreams?

In a remarkably short amount of time, the space has come to an answer: yes.

Yes… If You Have Action-Labeled Data (Or Can Get It)

Today, I want to share a little bit more about our approach and the results we’re starting to see.

Every approach I’ve written about so far runs into the same wall, eventually: It needs better data. Video is abundant, but it lacks depth. It has no action labels. And without knowing what actions caused what we’re seeing, video data is like shadows, the shadows on Plato’s cave wall.

And Yann may be right that you can infer action, but anyone using inferred action has separate scaling laws to consider: inferring the actions themselves. Inferring actions takes compute, time, and attention away from doing the things you can do once you understand actions, and while inferred actions might look good on benchmarks, they struggle deeply on edge cases. Even well-inferred actions are approximations of what someone actually did: some things just aren’t visible in video, like moving the rudder on a plane landing from the cockpit.

Hint: if you don’t do it, you crash. That’s why ground truth is crucial.

You need to find a way to get action-labeled data. The closer to ground truth, the better. Luckily, we have a fantastic starting point, thanks to Medal.

Before there was General Intuition, there was Medal

Earlier, we talked about the importance of games in the development of AI. AlphaGo. Deep Blue. These are intentional uses of games in AI.

There is an even richer history of accidental links between games and AI, lucky breaks.

Nvidia is the example you likely know. Jensen founded Nvidia to make chips for real-time graphics in games in 1993. Six years later, in 1999, Nvidia released its first “Graphics Processing Unit” (GPUs), the GeForce 256.

A few years later, around 2005, researchers began experimenting with GPUs for neural nets. In 2007, Nvidia released CUDA, to make ML on GPUs practical. In 2009, three Stanford researchers — Rajat Raina, Anand Madhaven, and Andrew Ng — showed that GPUs could accelerate deep learning by 70-100x for unsupervised learning.

Three years later, in 2012, the AlexNet team7 decimated their ImageNet competition using GPUs. Within a year, everyone in deep learning had switched to GPUs. “Everyone in deep learning” was still a small community at the time, but by then, Bitcoin miners had already been using GPUs. They were 50-100x more efficient for Bitcoin’s SHA-256 hashing than CPUs.

They soon switched to ASICs, but in 2015, Vitalik Buterin and his team released Ethereum, whose memory-heavy workloads were harder to optimize with ASICs. Ethereum mining ran on GPUs from 2015, through the GPU shortage it caused during the 2020-2022 crypto boom, right up until Ethereum switched from Proof-of-Work to Proof-of-Stake and left a GPU glut in its wake. Crypto tanked anyway, and in the same month crypto peaked, Nvidia’s stock did, too, tumbling 66% over the next year, until OpenAI released ChatGPT and, since then, Nvidia’s market cap has grown 10x to the $4.4 trillion behemoth we know today.

I mean, who could have predicted all of that?

When I taught myself to reverse engineer things, and learned how to code to build a private Runescape server when I was 13, I couldn’t have predicted that it would lead me to where I am today, either. Reverse engineering is the ultimate form of deductive reasoning, and spending a lot of time doing it as a kid is very good for your brain. This then lends itself well to figuring out complex systems in a rapidly changing world.

Runescape developers took the wilderness and free trade out of the game. I wanted to put it back, so I learned how to reverse engineer. The business that grew out of that did well for a teenager — we were making ~$1.5 million per year by the time I had to shut it down in 2015 at age 18, when I became an adult and would have been liable for the stuff I built. But I’d made enough money for my age that I could do whatever I was passionate about. I joined Doctors Without Borders (MSF) at 19 years old, and stayed for three years to work on Ebola & Humanitarian Mapping. I spent some time at Google Crisis Response before the gaming itch got me again.

At the time, we worked in London, very close to the DeepMind team. It was 2014 and I did not think it was that interesting or that likely to work. Demis deserves so much credit and respect for his vision. Few understand how hard it was for them to get here.

In 2018, I teamed up with my previous colleagues from building RuneScape servers. We built a game called Get Wrecked, which got a lot of signups. But it lost players quickly because we didn’t have enough player liquidity; it was a competitive game, and we needed to have enough people of all skill levels that people would always be able to find someone at their level to play against, which is very hard to bootstrap. To fix that problem, we built a way for people to watch game clips on the platform. A couple of times a day, we’d send a push notification that the game was live to get enough players playing at once.

The clips platform, Medal, went viral on the Rocket League subreddit. It was getting so many downloads that it became almost immediately clear that that was the bigger opportunity. We decided to focus on Medal.

We never ended up releasing the game. Medal just kept growing. Today, players from around the world upload 1B+ gaming clips to Medal each year.

We couldn’t have planned a better dataset with which to build World Models and policies.

Medal’s upload volume puts it on par with YouTube. Gamers upload millions of clips per day, across tens of thousands of environments, already hand-selected by players for highlights and adverse events. In other words, they share the content they think is worth sharing: their best performance, wildest encounters, closest calls.

Medal data has something that YouTube data does not. It comes enriched with metadata from our social network (views, likes, comments) and most importantly, in-game actions. We only record the game actions on the local machine, only storing the in-game action names (e.g. Move Forward) and never the keys that were pressed to achieve that action. More than data, this has enabled us to ship Medal’s most requested feature, keyboard and controller overlays. These overlays let our gamers showcase the precise actions they took behind every amazing moment.

Each clip has exactly what the player saw, next to the exact player actions that followed, using many of the same systems we use to control robots today. Frames from games also have the benefit of being information-complete. Unlike real world video, where you have to account for pose estimation (estimating what the human sees, which in itself is a lossy process) — you may see things the camera does not in the real world — but not in games. What is recorded and what you see is always identical, which we think makes it better training material.

This gives us trillions of examples of players running the loop of observe, predict, and act. This is the foundation of intelligence, and there is no loss of information throughout.

On Data

To understand what we’re doing, you need to grok the difference between game data and synthetic data.

The confusion is that people associate “digital” with “synthetic,” but the real distinction isn’t the environment in which the data was generated, but the data itself.

There can be synthetic (i.e., generated) data created in the physical world, like in the human-constructed environments Boston Dynamics and other robotics companies train their robots in, just as there can be human ground truth data in the digital world. Data breaks down into a quadrant like this.

What makes our gaming data “ground truth human data in a digital environment” is that what we are capturing is real human responses that observe → predict → act loop.

The closest comparison to our approach is GitHub data. It captured the history of human engineers’ coding and was used to train machines that can code better than humans. The question is whether that same idea works outside of the computer. We believe (and are seeing signs) that learning from gaming data transfers to the physical world.

Games turn out to be the perfect training ground for learning intelligence. They contain thousands of simulated worlds with physics, strategy, cooperation, text, interface use, competition, and long-horizon planning. They are complex enough to require intuition, but structured enough to learn from at a massive scale.

Physical-world data alone cannot reach the diversity or scale required to learn general intelligence. LLMs lack data on dynamics and atoms. But games act as the ideal intermediary: a bridge between the digital world of bits and the physical world of atoms.

Still, there is a threat to the ground truth stance. As mentioned earlier regarding Yann LeCun’s take, every video is action-labeled data if you’re good enough at inferring action. While this may be true in the long run, it’s also probably extremely impractical today. That’s why you gotta love Yann — nobody else would think to do it this way. Yann and I talked about this dilemma in Paris in December if you want to go deeper into the weeds.

Everything is a trade-off, right?

The optimal path forward is likely somewhere between where VLAs are today — the most practical but least elegant solution — and where AMI might be one day if everything goes well. It all comes down to your approach to data.

Data is the problem for any company that wants to solve embodied AI. Evan and Packy wrote about it in Many Small Steps for Robots, and it is the thing we are focused on at GI.

We believe our dataset is the most elegant answer to the data problem for general models. It is one that paves a path towards a general intelligence that feels familiar, the same way Tesla FSD feels like a familiar driver, but scales far beyond games or driving.

For general models, models that can power embodied AI intuitively and spontaneously in almost any imaginable real-world situation, the question is not simply how much data you can get.

Before throwing data at the problem, you need to understand your transfer curves.

Small Steps, Giant Leaps, and Transfer Curves

In their Robot essay, Packy and Evan wrote that there are two approaches to building economically viable embodied AI: Small Step or Giant Leap.

Evan and his company Standard Bots are pursuing the small step approach: getting paid to learn in the field, one use case at a time. They are collecting real-world data for a growing number of economically viable use cases across a wide variety of domains.

Their strategy is a fascinating one. By getting customers across many different industries and tasks to pay them to deploy, they collect diverse real-world data across wide distributions. Instead of hoping that more data in a narrow domain will generalize to tasks outside of the distribution, their goal is to put a ton of useful tasks in-distribution by going broad in the real world, not deep in one niche.

General Intuition and Standard Bots are coming at the same problem from opposite sides of the spectrum. General Intuition is attempting to solve generalization from the digital side: our bet is that gaming data will lead to broad priors about physics and actions. Standard Bots is attempting to solve generalization from the physical side: their bet is that real-world deployments will lead to broad priors about manipulation and industrial tasks.

These are complementary approaches to the data diversity problem. There’s potential for a GI World Model to be the starting point for Standard Bots’ post-training. We provide a base model trained on observed data in digital environments, which is scalable and economical to collect, and they post-train using the use case-specific data they get paid to collect to bring those use cases in-distribution and get to many 9s of reliability more quickly.

The approach that we find more challenging is the one that general models seem to be taking, which is collecting a ton of data and hoping it generalizes to out-of-distribution tasks. General models need too much data across too many situations to collect it all by paying people to demonstrate tasks.

Additionally, more data in the same domain doesn’t automatically teach a model to handle situations it’s never seen before. At pre-training, not all data is created equal, and I have not met someone building a general model for robotics who has been able to point me to the scaling laws that show that they can solve out-of-distribution use cases (things they weren’t trained on) simply by adding more data. More data in a narrow domain does not automatically buy you generalization to a new one.

The scaling laws don’t exist.

There are, as best we can tell, three distinct transfer curves that govern whether a World Model generalizes to new physical environments. They are not well understood — we are just starting to understand them. We can, however, give them names: input modality transfer, sensor transfer, and environment transfer.

The first is input modality transfer: How well does a policy generalize across degrees of freedom in the physical system it’s controlling? This curve is steep for a humanoid robot with somewhere between twenty and sixty degrees of freedom. Each is continuous and often mechanically dependent on the others, and this curve is steep. A finger movement is not independent of the arm. Training on a game controller and expecting it to transfer cleanly to a twenty-DOF humanoid hand is, in research terms, a bet without scaling laws to back it up.

The second is sensor transfer: If the workload requires specialized physical sensors (tactile feedback, proprioception, depth), there is a separate scaling law for how much of that sensor-specific data you need before the model can reliably reason about it. Tesla explicitly worked on this problem. They spent years figuring out exactly how much LiDAR data they needed before they could entirely drop the chips. Most robotics companies are working on it implicitly, hoping the answer reveals itself in deployment.

The third is environment transfer: How does performance degrade as the environment gets more complex, more stochastic, or more populated? Predicting the right action in a sports stadium with a thousand people around you is a fundamentally harder problem than predicting it on an empty field.

As we explained earlier, complexity doesn’t scale linearly.

These three curves interact. Until you can map them, you can’t know how much data of which type you actually need, which means you cannot justify the capital expense of going to collect it at scale. Companies that are collecting a hundred thousand hours of physical data today may find that a good World Model only needed ten thousand, or that they did need a hundred thousand, but ninety thousand of the hours they got were in the wrong distribution entirely.

Our bet, certainly conditioned on our starting position, is to collapse the problem.

By focusing on game controller inputs, we reduce input modality transfer to a curve we have already solved. We know we have enough data for game controllers, because we have billions of clips of humans using them. That eliminates one unknown. By focusing on vision-based inputs rather than specialized sensors, we eliminate the second.

Almost every physical system comes with a game-controller-like input modality, which includes steering wheels, keyboard and mouse and actual game controllers. Most are straightforward. Even humanoids ship with them. The challenge is just that if the degrees of freedom exceed what the controller can do, transfer is worse. So humanoids are further down our roadmap, but we see no physical limit to suggest that we couldn’t build around the interface limit.

In short: If you can control almost any physical system with a game controller, and we have more data than anyone else in the world of what happens when a player uses a controller to take action, our Agents should be able to control almost any physical system.

The only remaining question is about environment transfer: can Agents trained inside of the dream operate in reality?

The Superhuman Future of the World (Models)

It has been a wild few weeks at General Intuition’s offices in New York and Geneva. Everything that we’ve written about here has been working better than we expected. Like others, we’re gaining conviction that Agents trained inside of the dream can operate within reality.

Why do World Models transfer?

The observe-predict-act loop is an abstraction for how causally structured systems work in general. Once a world model has seen N variations of the world via a diverse set of games, it only takes a small amount of fine-tuning to understand the dynamics of the N+1 variant that corresponds to the real world.

World Models learn to model the cause-and-effect of reality. If this cause-and-effect is understood at a fundamental enough level, this should enable World Models to generalize to new scenarios.

What might that mean? What are the implications of World Models that generalize?

Our goal is to enable embodied AI to understand the world, with our models controlling machines in any environment, including the real world. We aim to deliver a breakthrough moment for robotics, where out of nowhere, the progress is obvious and the models are easy to use.

That breakthrough won’t look like the breakthroughs in LLMs, which went mainstream when they started talking to us like humans. We don’t want machines that simply do what humans do. In fact, the point of machines is to do things that humans can’t, to give us superpowers.

Robots don’t need to look like us to work for us. Humanoids as a form factor were largely chosen due to the assumption that they had the most data to learn from on the internet, because so many of humanity’s videos feature humans. If you don’t need those videos, if you can learn directly from the actions in video games across embodiments, and need a lot less to transfer to reality, that assumption doesn’t hold. We believe the future of robotics should be shaped by simpler, cheaper systems: machines with only the degrees of freedom that match the actual jobs to be done.

The human body is an incredible general-purpose platform, but it’s rarely the optimal (or most cost-effective) form for any specific task. Instead of copying our anatomy, we should mirror the interfaces we already use instinctively: joysticks, wheels, gamepads, and keyboards. These tools are the product of decades of iteration, compressing human intent into a clean, universal action space, much like language does for thought. Robots can learn from the actions transmitted through these interfaces and specialize around them in a very general way, making broad deployment far more practical than chasing full human embodiment.

If you get rid of the assumption that our machines don’t need to, and probably shouldn’t, mimic us or take our place in any way, a whole world of possibilities opens up.

At General Intuition, we’re actively working on simulations that will eventually allow our systems to go beyond everything that’s currently described in pixels, to everything governed by cause and effect. The methods we use are very general. This is a long way out, but a necessary step.

To really understand our world, it turns out, poetically, we may need World Models; compute for the uncomputable.

The implications of all of this are cosmic. If we can model three-dimensionality, physics, and time, and their interaction, then the ability to manipulate these arenas at superhuman macro and micro scales is on the horizon.

There is a tremendous amount of work ahead. Today, nobody is capable of simulating a biological cell, let alone an ecosystem made up of 10³⁰ of them. However, what captivates me is that we don’t need to map all of reality’s details. We just need to observe how those details manifest in actions, and use those actions to predict what comes next, over and over again.

There is also a tremendous amount of responsibility that comes with building these models, and it’s something I take very seriously and personally.

I’m from the generation most at risk of AI displacement; half my childhood friends can’t find jobs. I’m spending a lot of time exploring how we bring our community and my generation along in this shift.

For example, like Tesla, Medal sits on over $10 billion of global hardware infrastructure — GPUs, CPUs, plugged into power, with cooling — powered by over 15 million users. We’re actively exploring ways to let our community share in what’s coming, for example by generating income serving inference from their GPUs, or tele-operating from their gaming rigs. If the demand for general intelligence is anywhere close to what we think, that could be the largest economic tailwind my generation has ever seen.

These are just my dreams for now. But one day, they won’t be. One day, we leave the boring problems to superintelligence, so we can explore the stars or the deep sea from our gaming rigs, and dream up the next uniquely human, most interesting, not boring things to do.

If our work interests you, I’m recruiting to work on simulators to generate new environments, and we continue to be excited to talk to researchers and engineers who are in the top of their field and want to join General Intuition.

A special thanks to Eloi Alonso, Adam Jelley, Vincent Micheli, and Paula Wehmeyer, my co-founders and colleagues who spent many hours discussing the ideas behind this article. - Pim

Big thanks to Pim, Paula, Adam, Eloi, Vincent, Kent, and the whole General Intuition team for sharing their knowledge, and to Badal for the cover art. - Packy

That’s all for today.

For not boring world members, I played around with Claude to produce some extra goodies. We made a World Model Research Archive, with links to more than 30 key papers that have defined the space. Members can also ask Pim questions in the subscriber chat today and tomorrow.

Join us in not boring world for all of this and more by subscribing below

Thanks for reading,

Packy

Subscribe now

not boring world members can download the World Model Research Archive here.

Hi friends 👋,

Happy Friday! Here in New York City, it was the sunniest of times, it was the snowiest of times, but most of all, it was the incredibly hard to keep up with all of the incredible stuff getting done and funded of times.

Big week in robotics, solar, brains, world models, manufacturing, flight, AI, and even affordable luxury.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Ramp Sheets

A few weeks ago, I wrote that I wrote that “a Ramp engineer with an AI will build something better than a CFO with an AI.” Ramp Sheets is exactly what I’m talking about.

See, a surprising amount of my high-leverage work ends up in a spreadsheet.

A deck comes in. A CSV export lands in my inbox. I want to benchmark a market, pressure-test a company, or run an analysis for a Deep. Dive. Usually, that means a lot of manual cleanup and model-building before I get to the actual insight.

Ramp Sheets skips a lot of that.

It’s an AI-powered spreadsheet from Ramp that lets you upload a PDF, pitch deck, CSV, or Excel file and ask it to do the work: build a model, research competitors, clean messy data, and help you get to an answer faster. For founders, investors, and operators, that means less spreadsheet mechanics and more actual thinking.

Best part: anyone can use Ramp Sheets today for free. You don’t need to be a Ramp customer to try it (but why wouldn’t you be?).

Try Ramp Sheets Free

(1) Swift Solar Acquires Meyer Burger to Build Gigawatt-Scale Solar Factory

for

“If your technology is so good, why aren’t you using it to compete?”

In Power in the Age of Intelligence, I wrote that that’s a question I’d like to see more investors ask. If the technology you’re building with is so much better, why in the world are you just selling it to someone else as a component? Why not just make a better end product?

Joel Jean at Swift Solar said, “Ok bet.”

I met Joel a few years back on an introduction from his brother Neal, the co-founder and CEO of Beacons. Joel, like Neal, is a genius, with a PhD in electrical engineering from MIT, and he was putting that genius to work by making silicon-perovskite tandem solar cells work.

The promise of perovskite is that they’re much more efficient than solar panels, meaning they can convert more incoming sunlight into electricity. Current commercial silicon panels are 22% efficient, with a lab record of 26.8% and a theoretical limit of 29%. Swift Solar’s panels are already at 28% efficiency. China’s Longi, the world’s largest manufacturer of monocrystalline silicon wafers, recently hit 34.8% with a silicon-perovskite tandem solar cell, and the theoretical limit, for which Swift is gunning, is 45%.

More efficiency means more watts per panel, which means more watts per acre and a lower balance-of-system cost; 30–40% efficiency modules could cut solar system costs 20–40% while doubling power density.

This all sounds great, but the problem is that tandems tend to degrade quickly outside, where the sun is. Swift thinks that they’ve cracked that, and that China hasn’t yet, which gives the west an opportunity (certainly not guaranteed, probably not even likely, but an opportunity!) to leapfrog back over China in solar production.

So if you have the magic, degradation-resistant tandem cells, what do you do? Use it to compete.

Swift announced that it acquired the core assets of Europe’s leading solar manufacturer, the German company Meyer Burger, to “build the next generation vertically integrated US solar manufacturer.” Meyer Burger’s former CEO joined Swift to lead the manufacturing effort.

Solar panel manufacturing is one of the most cutthroat, competitive industries in the world. China keeps driving costs lower and lower and lower. It’s been all about scale. Meyer Burger itself went bankrupt after having sold solar cell manufacturing equipment to China, only for China to reverse engineer and mimic the technology.

But I love it. Pressing a technological advantage into vertical integration and competing directly is the kind of thing that gets my heart racing. And this echoes a deal that hits close to home. A few weeks into Puja’s time at Harry’s, she told me the 10-month-old startup was acquiring one of the world’s oldest and best blade manufacturers, Germany’s Feintechnik. That has been an incredibly successful partnership.

There’s something about a the US innovation / German manufacturing tandem.

(2) Reflect Orbital Built a Space Mirror

Ben Nowack for Reflect Orbital

Imagine you are a photon that does not want to get turned into usable electricity on earth. It’s been a really bad week for you.

First, Swift’s tandem cells. Before, there was a 78% chance you’d escape once you hit the panel, but that looks like it’s going to fall. But at least you have nighttime, right?

At night, you can just fly off into space at the speed of… well, you know. Run away from Earth as fast as is physically possible.

Not so fast, says Reflect Orbital. CEO Ben Nowack just announced that the company had made a solar mirror, one of four huge guys that will go up on each satellite, stop the photons’ getaway, and redirect them down to Earth.

Sometimes, they’ll be used to light up a concert or a remote worksite in the evening (imagine how much less creepy True Detective: Night Country would have been if Reflect existed then), sometimes, they’ll shine on crops to help them grow faster, and mostly, they’ll bounce into solar panels that capture them and turn them into electricity.

Efficiency is one of solar’s challenges, but a bigger one is one that seemed almot insurmountable: the sun only really shines strongly enough for max solar output for 4-6 hours a day. What if it could shine on solar farms - out in the desert, away from homes (where it wouldn’t imapact our sleep, Huberman) - round the clock?

Solar farm economics would get ridiculous, solar would get much cheaper, we’d get a lot more of it, and we’d be able to power all sorts of new, better electric things.

Mostly, though, I’m sharing because that video is a bright spot in a sea of sameness.

(3) The First Multi-Behavior Brain Upload

Dr. Alex Wissner-Gross for Eon Systems

As big a week as it was for solar, it may have been an even bigger week for brains.

A couple of weeks ago, an Australian company Cortical Labs grew brain cells on a microchip and taught it to do the first thing hackers try to do on any new computer, from TI-83s to ATMs to pregnancy tests to vape screens: play Doom. And yup, it could play Doom.

On Tuesday, Dr. Alex Wissner-Gross shared a video made by a company he works with called Eon Systems, that one above, that shows a virtual fruit fly being controlled by a brain with 140,000 neurons and 50 million synaptic connections running on a computer, mapped from a real fly’s actual connectome.

The Eon team took the adult Drosophila brain connectome, the complete wiring diagram of a real fly brain, painstakingly mapped by the broader neuroscience community, combined it with a physics-simulated fly body built from an X-ray scan of an actual fruit fly, and got the two to talk to each other. Then sensory information from the virtual world enters the brain model, the brain’s neurons issue commands, and the body moves. It does all of the things a fruit fly would do in response to stimuli. It navigates toward food by “taste.” It grooms itself when virtual dust activates its “antennal mechanosensory circuits.” It “eats.”

The team is careful not to oversell the accomplishment, and Dr. Doris Tsao at Astera Neuro has a good list of thoughts on connectonomics and uploads here which basically says, “This is awesome, but we’ll get further by understanding instead of blindly mapping, and anyway, it’s going to be really expensive to map bigger and more complex brains.”

But still… this is sci-fi stuff. I don’t even know if it’s good. But it’s sci-fi!

Meanwhile, in regular old science, two very cool brain papers:

1. Isotonic and minimally invasive optical clearing media for live cell imaging ex vivo and in vivo. A team of Japanese scientists basically figured out a way to “make brains see-through without killing them.” Kording Lab explained: “The mechanism of this is so obvious and simple and yet brilliant. Match the refractive index inside and outside of cells - and tissue becomes transparent. Because the scaling of scattering with object size, small things don't matter much. So cool!”

2. Scientists revive activity in frozen mouse brains for the first time. A team in Germany made progress on cryopreserving and thawing mouse brains that leaves some of the processes necessary for brain functioning like neuronal firing, cell metabolism, and brain plasticity intact. A researcher at another lab cautioned that we’re still a long way away from freezing ourselves and waking up in the future, but he also said, “This kind of progress is what gradually turns science fiction into scientific possibility.”

OK, so maybe all of this is just very sci-fi.

(4) Emergent Quantization from a dynamic vacuum, or Zero Point Energy

Harold “Sonny” White et al, published in APS Physical Review Research, via Andrew Côté

Maybe my favorite rabbit hole I’ve fallen down over the past few years is the Zero-Point Energy Rabbit Hole, which I discovered via .

The idea is that at absolute zero, classical physics says everything should stop moving, but the Heisenberg uncertainty principle prevents any particle from simultaneously having a precise position and zero momentum, so it has to keep jiggling. That constant jitter is ZPE.

The vacuum, which we think of as empty, is a seething foam of energy, with virtual particles constantly popping into and out of existence. This sounds like woo, but it’s been experimentally confirmed via the Casimir effect: two uncharged metal plates placed extremely close together in a vacuum attract each other, because ZPE fluctuations are slightly suppressed between the plates compared to outside, creating a net inward pressure. This has been measured in real labs like Los Alamos.

Quantum theory predicts, and experiments verify, that empty space contains an enormous amount of ZPE. The amount of energy supposedly packed into the vacuum would constitute a seemingly ubiquitous energy supply, a “Holy Grail” energy source, which is really just about as close to a “Holy Grail” as we can get. You can do a lot with unlimited energy.

Stuff gets wilder from there in a chain of logic that leads from ZPE to mass, inertia, and gravity as engineerable properties of the vacuum to the ability to actually engineer the vacuum to create phenomena like antigravity, which is what some believe Thomas Townsend Brown did.

Anyway, a team led by Dr. Harold “Sonny” White, a physicist and aerospace engineer with experience at Boeing, Lockheed Martin, and NASA who is now the CEO of ZPE company Casimir, showed that “Instead of random noise, the quantum vacuum can have structure, meaning waves, dispersion, resonances,” per Andrew Côté, who added that “their analytical model reproduces all the 'larger scale' predictions and measurements of quantum mechanics exactly.”

Andrew said that while the difference sounds subtle, “it means the vacuum can have local structure, that it is in effect a dynamic medium,” which means that, potentially, we can pull energy out of it. That is what Casimir is trying to do.

Now look, Casey Handmer replied betting $1 million that it wouldn’t work, and who am I to bet against Casey Handmer.

But 1) there are a lot of smart people on the other side of that bet, like Andrew, Sonny, Hal Puthoff, etc… and 2) even if there’s a 1% chance there’s something there, it’s hard to think of anything bigger.

(5) Replit Launches Agent 4

Amjad Masad for Replit

Look, I’m biased. I’m an investor in Replit. I wrote about them in 2021 and predicted they’d be a $100 billion company. This week, they raised at a $9 billion valuation, so we’re getting there.

More importantly, though, they launched their new product, Agent 4. While I love Replit, and think Amjad is one of the best CEOs out there, I’ve been a little worried about vibe coding as a category. It’s so competitive, and won’t the big labs just do it? Won’t Claude Code kill everyone else?

Maybe. Certainly a possibility. But you should watch this video. By focusing maniacally on this one software creation product, the same thing they’ve been focused on making easy, from the infrastructure up, for over a decade, before AI was even a thing, they’ve built what I think is a truly differentiated product, and one that feels like one of the first good AI-native products I’ve seen.

You can collaborate in teams, set multiple agents to work at once, workshop designs, give feedback on specific ones, tweak them, iterate until you get something good, even doodle to show the AI what you want, and when you’ve found a design you like, you can keep it consistent across website, mobile app, presentations, docs, and more. It feels like a workspace for teams, where some of the teammates, the ones doing most of the grunt work at the humans’ direction, happen to be agents.

I’m going to be playing with it this weekend. If you have some time, build something.

EXTRA DOSE: Enormous Week of Fundraising + Nietzsche as Mystic for not boring world subscribers (cmon, join us)

Read more

Hi friends 👋,

Happy Friday and welcome back to our 183rd Weekly Dose of Optimism.

I was at a16z’s American Dynamism event in DC this week, where I got to meet a bunch of people who read not boring, including one who told me she never opens the Weekly Dose because the title doesn’t tell her anything.

If you’re reading this: thanks for the feedback, I guess it worked. I’m running a title test; I’ll check open rates this week and decide what to do going forward.

Those of you who opened are in for a treat. What a week for the optimists.

Let’s get to it.

Today’s Weekly Dose is brought to you by… Guru

Guru is the best way to make sure that your employees get accurate company knowledge from whatever AI you use.

Companies like Spotify and Brex use Guru because it’s the only AI verification system that automatically validates company knowledge before your AI agents use it, like quality control for your AI’s brain.

Like them, your company’s systems weren’t built to power automated systems. Your people send Slack messages, write docs, and even quasi-maintain wikis for other people to consume. Guru translates all of it to speak AI so that AI doesn’t give your people incorrect answers ~40% of the time.

Put your messy company data to clean use in a couple of clicks. Get a Guru to do it for you.

Try Guru Today

(1) a16z American Dynamism Summit

On Tuesday, I Joe Bidened myself1 down to Washington, DC for a16z’s American Dynamism Summit.

Complete with its very own Jason Carman hype video (see above), it was like a living, breathing version of the Weekly Dose of Optimism, and full of the people behind the stories we cover here each week.

Drew Baglino, whose Heron Power led off the Dose last week, was there, as was Noah Smith, whose Decade of the Battery I linked in that story. Cameron McCord, whose fundraise for Nominal you’ll read about today, was there, too. I spent time with probably a dozen people whose work has appeared in the Dose, and got to watch others on stage, like NASA Administrator Jared Isaacman.

My job is to spend time with these people, write about them, and invest in some of their companies, but there was something about seeing so many of them in one place that brought home the fact that a small group of sufficiently talented and motivated people can change how the future plays out.

Power generation, transmission, and storage, motors, power electronics, autonomous boats, supersonic planes, educational choice, modern manufacturing (terrestrial and in space), new weapons systems, cheap and secure communication everywhere, autonomous everything, safer neighborhoods, new cities altogether. If the people in that room succeed, America’s next 250 years will be better than its first 250.

(2) Physical Intelligence Models Get Memory So Robots Can Do Complex Tasks

Last week, I was talking to a Physical Intelligence investor who told me the progress the company had made in just the first couple of months of 2026 has blown his mind. Now, we know why.

Pi gave its models a new memory system that helps both short-term visual memory, to recall what it did recently with fine-grained details, and long-term semantic memory, which lets it remember what it did for up to 15 minutes.

With better memory, robots using Pi’s model can do long, multi-step things, like clean up an entire kitchen, set up the ingredients for a recipe, and grill a grilled cheese sandwich. It also learns from its past mistakes.

Yesterday, I wrote that LLMs are producing a lot of dead text that we’re probably better without. Robotics is way more interesting. I would love for the robots to spend billions of tokens doing my laundry and making me healthy gourmet meals while I kick back and read human-written classics. This is one more small step in that direction.

For a full primer on robotics, check out my cossay with Evan Beard:

(3) Arc Institute Releases Evo 2, Fully Open-Source Biological Foundation Model

A year ago, the Arc Institute released Evo 2, the largest fully open biological AI model ever built, as a preprint. This week it was published in Nature, and Arc dropped a one-year recap showing what the thing has actually done in the wild.

Trained on 9.3 trillion nucleotides from over 128,000 whole genomes, Evo 2 can read and write across all three domains of life: bacteria, archaea, and complex eukaryotes including us. The architecture, called StripedHyena 2, trained on 30 times more data than its predecessor Evo 1, and can reason over 8 times as many nucleotides at a time. Greg Brockman, the OpenAI co-founder and ex-Stripe CTO (Stripe co-founder Patrick Collison is a founder of Arc), helped build it during a sabbatical.

Evo 2 generalizes “across biological prediction and design tasks, across all modalities of the central dogma, across molecular to genome scale, and across all domains of life.”

It can predict which genetic variants cause disease and which don’t, without being specifically trained on that task. In tests on the breast cancer gene BRCA1, Evo 2 achieved greater than 90% accuracy in predicting which mutations are benign versus disease-causing.

It can generate new genomes. Entire genomes. The team demonstrated the first AI-designed and experimentally validated bacteriophage: 16 of 285 tested designs successfully propagated and inhibited growth of the target bacteria, with no impact on unrelated strains. For those who, like me, don’t fully speak the language of biology, that’s a whole functional new organism.

And in a stunt that’s both scientifically real and deeply fun: the team guided Evo 2 to design DNA sequences with controllable chromatin accessibility patterns, then wrote Morse code messages—”EVO2,” “LO,” and “ARC”—into the epigenome of mouse embryonic stem cells, experimentally validated with ATAC-seq.

The whole thing is fully open: model weights, training code, inference code, and the OpenGenome2 dataset. So more scientists will be able to experiment with a model that understands the grammar of life as the cost of reading and writing biology keeps falling. The design space for medicine, agriculture, and materials is going to keep opening up in ways that are genuinely hard to imagine.

(4) Injectable “satellite livers” could offer an alternative to liver transplantation

Sangeeta Bhatia Lab / MIT / Cell Biomaterials

More than 10,000 Americans are on a waitlist for a liver transplant. There aren’t enough donated organs. And many patients with liver failure aren’t even eligible for the surgery because they’re too sick to survive it. For those people, the best current answer is: sorry.

MIT engineer Sangeeta Bhatia’s answer is: what if we just injected a second liver?

Her lab has developed “satellite livers,” tiny engineered tissue grafts made of hepatocytes (the liver's workhorse cells) mixed with hydrogel microspheres and supportive fibroblasts, all deliverable through a syringe. Once injected into fatty tissue in the abdomen, the spheres act like a liquid going in and then solidify, giving the hepatocytes a scaffold to cluster around. Blood vessels grow in and the mini liver takes root. In mouse studies, the grafts remained viable and producing the right proteins for at least eight weeks (the full length of the study).

The insight that makes this work is a clever one: the hydrogel microspheres create an “engineered niche” for the cells. Without them, injected hepatocytes disperse and fail to integrate. With them, the cells localize, connect to the host circulation faster, and behave the way liver cells are supposed to.

The vision is a twofer. Short-term, these grafts could act as a bridge to transplantation, buying time for patients waiting on the donor list. Long-term, they could be the treatment: repeatedly administered, minimally invasive, titrated to how much liver function a patient needs. It could replace surgery altogether with routine injections.

The liver does about 500 essential things for the human body. The transplant chain is broken. People die. The fact that we might soon be able to restore meaningful liver function with a syringe, a handful of cells, and some hydrogel spheres is miraculous.

(5) Ginkgo Bioworks Launches Ginkgo Cloud Lab

In early February, we shared Ginkgo’s collaboration with OpenAI in which GPT-5 integrated with Ginkgo’s autonomous lab and autonomously designed, executed, and analyzed over 36,000 cell-free protein synthesis experiments. GPT-5 reduced costs by 40% compared to benchmarks (from $698 to $422 per gram). Good robot.

Now, Ginkgo is making its Cloud Labs available to anyone in a push to move all of its R&D services away from traditional benches and to these futuristic ones. Building out a lab can cost a bunch of money, Ginkgo CEO Jason Kelly tweeted that it can cost $10 million, and Ginkgo’s strategy has long been to offer lab services to startups, academic labs, and pharma companies as a service.

We hope that Cloud Labs will someday allow anyone to be a scientist with their own lab just like personal computers and cloud data centers democratized programming and the web.

With this move, presumably the service gets better as Ginkgo’s margins do, too. The starting price for Ginkgo’s first three protocols, two for cell-free protein expression and one for bacterial pixel art, is just $39.

Awesome news for the democratization of science, although I can’t imagine this is going to help my friend sleep any better at night.

Noahpinion

Updated thoughts on AI risk

So the other day I wrote a post about how humanity is inevitably going to be disempowered by the existence of AI…

Read more

2 months ago · 408 likes · 100 comments · Noah Smith

EXTRA DOSE: Ulkar is out for the week, so in our Extra Dose, I’m rounding up a big week in fundraising news for software for hardware and sight restoration, plus recommending my favorite sci-fi series of all time.

Read more

Subscribe now

Hi friends 👋,

Happy Thursday! It’s late in the week and late in the morning, so…

Let’s get to it.

Today’s Not Boring is brought to you by… Deel

Every week, at the bottom of the Weekly Dose, I thank my teammates, Aman and Sehaj, for their contributions. Thanks isn’t enough though. They do great work, for which they deserve to get paid. Challenge is: they’re based in India. Or, challenge was. Now, I pay them with Deel.

Deel is payroll for global teams, like not boring and many of our portfolio companies. I grabbed coffee with a founder in our portfolio the other day, and unprompted, he brought up how easy Deel made it to expand his team globally. He’d tried the alternatives and switched, and he was gushing, which is rare for a payroll provider.

Hiring talent outside your home country can open up new growth, but it also comes with unfamiliar rules, local labor laws, and compliance risks. That’s where an Employer of Record (EOR) comes in. If you’re a startup exploring global hiring for the first time, Deel’s free guide walks you through the basics. Whether you’re hiring your first international employee or planning your next market expansion, Deel can help.

Download the Guide

Costless Sacrifice

The Old Testament’s 2 Samuel tells the tale of King David as he unifies the tribes of Israel and establishes Jerusalem as the nation’s capital. The world was slower then, there were only like 50 to 75 million people around at the time, and so God could be more actively involved in the day-to-day management of human affairs.

As, for example, when King David ordered a census of Israel to put a number on its military strength, he angered God, who expected his servant to rely not on soldiers but on Him. God sent a plague down that killed 70,000 men, then instructed David, through the prophet Gad, to build an altar on the threshing floor of Araunah the Jebusite in order to stop it.

When David arrived, Araunah offered to just give him the land, the oxen, and the wood with which to sacrifice them for free as a gift to the king.

“But the king replied to Araunah, ‘No, I insist on paying you for it. I will not sacrifice to the Lord my God burnt offerings that cost me nothing.’ So David bought the threshing floor and the oxen and paid fifty shekels of silver for them.”

Costless sacrifice is not sacrifice. And we, like the Gods, demand sacrifice.

I read an essay in The Argument by economist yesterday: The Tinder-ization of the Job Market.

The Argument

The Tinder-ization of the job market

Is AI going to take everyone’s jobs? A recent report went viral after claiming there were 55,000 AI-related layoffs in 2025…

Read more

2 months ago · 69 likes · 19 comments · Matt Darling

The job market is stuck, Darling argues. It’s not bad: unemployment was relatively low in January at 4.3%; Prime Age (25-54) Employment at 80.9% is higher than it was at any point in the Obama or first Trump presidencies. Just stuck. The hiring rate averaged 3.3% in H2 2025, a level only touched during COVID and the Global Financial Crisis.

BLS Hires Data

The weird part is… we’re not in the middle of a crisis, financial or biological. Employment is high, remember! Unemployment is low! “Generally, a high employment rate and a ‘tight’ labor market are associated with high hiring rates,” Darling writes, “not low ones.”

Darling’s theory is that LLMs have made it easier for people to apply for a ton of jobs, with custom-written cover letters and everything.

He cites Greenhouse data which “showed recruiting workload rose 26% in the third quarter of 2024 and that 38% of job seekers reported ‘mass applying,’ flooding firms with far more resumes than before.” Per Business Insider, “the applications-to-recruiter ratio is now about 500-1, four times what it was just four years ago.” He links to a Kelsey Piper essay from last year that provides more data to back this up.

Because it’s easier to apply, the volume of applications is up. Because AI is writing cover letters, their quality is no longer a useful way to pull signal from all that noise. “A recent paper showed that after Freelancer.com introduced an AI-generated cover letter tool, the correlation between cover letter customization and offers dropped 79%.”

You used to get rewarded for customizing your cover letter, when it cost you something. Now it doesn’t, so you don’t, but you still gotta customize, because everyone else is.

We find ourselves running an incredibly stupid Red Queen’s Race.

“Here’s the hard thing about easy things: if everyone can do something, there’s no advantage to doing it, but you still have to do it anyway just to keep up.” I wrote that way back in August 2020, about DTC brands and Shopify. “When every rebel is armed, none really is. It’s like when you played GoldenEye 007 as a kid. Getting the Golden Gun the hard way was dope. Everyone getting the Golden Gun with a cheat code made the game suck.”

This is an idea I’ve been obsessed with for a long time: the asymmetric ability of the laggards to make the leaders’ life a little bit worse.

You might produce the very best widget by far, but your potential customers will undoubtedly be bombarded by inferior alternatives claiming that they make the very best widget. At worst, your potential customer will fall for it; at best, your cost to acquire that customer goes up a little bit.

You spend hour after quiet hour handcrafting your widgets, smoothing their edges, polishing their faces, giving each one a little kiss before sending it out. You sacrifice. Your competitors do none of that. They yeet millions of these suckers out in Yiwu. Then they stamp HANDCRAFTED WITH LOVE IN USA on their website and who’s to say?

Some people really want a particular job. It’s their dream job. They spend hour after quiet hour poring over resume details, crafting a heartfelt cover letter, and saying a little prayer before hitting “Submit.” They sacrifice. Their sacrifice gets lost among the 1,732 less-lovingly-crafted (but who’s to say? who has the time to read them all) applications that came in that same day.

The burnt offerings that cost nothing and the burnt offerings that cost everything smell the same, because we make our offerings not to an omniscient God, but to fallible, overwhelmed humans.

This is happening more and more with writing, too. Claude in particular has gotten much better at writing, which means that more people are publishing essays.

There will be times when there’s an interesting topic to write about, and by the time I’ve thought about what path I might take through the argument, there are dozens of versions of the essay on X. Some of them are even pretty good!

But I’ve noticed something happening in my brain that’s worth sharing, because I would imagine a lot of people are thinking similar things about whatever it is that they do.

I’ll be interested in a topic, start chewing on it, start thinking through unique ways to present it, to shape the essay, the research I might need to do, the people I might need to talk to, all this stuff I’d need to do to make something great. Then I see a handful of pretty good versions on similar-ish topics, and I think, “Well, I guess they beat me to it. On to the next one.”

And who knows if the version I would have written would have ended up being better than what these AI-human teams whipped up in a couple hours. Who knows if it would have been great, or even good.

What I do know is that if I’d written it, I would have put in a lot more effort, agonized over it, sweat the details, tried to present the ideas in unexpected ways. I would have sacrificed something that cost me something, hours and hours, sometimes weeks and weeks of my life.

And half of the value of the post would have been the result of that effort, but half would have been the effort itself, my pointing to an idea and saying, “THIS IS AN IDEA THAT I WAS WILLING TO SPEND FIFTY HOURS WRESTLING WITH.”

And maybe I’ll still spend the time to write it, and maybe many someone elses will spend the time making the full and beautiful version of the things they want to create despite the easy-come existence of their bloodless simulacra, but a lot of the time, we won’t, and the world will be left with many hollow versions of a thing filling up the place where the one full one should have gone.

The pretty good is the enemy of the potentially great.

Right, like someone will somehow, eventually, once the recruiters have gathered themselves up and faced the never-ending application pile, basically just win the lottery and get offered the job, and they’ll be happy that they got it, because a paycheck is a paycheck, but two things will probably be true: 1) it wasn’t their dream job; they applied to 347 of them, this was the one they got, and 2) the person whose dream job it actually was, who would have been overjoyed to get to work in this specific job, who actually did write the cover letter the old-fashioned way for this one because they reallllllllly wanted it… no human ever even saw that person’s application. They’re still looking. Dream crushed, they’ve now applied to 123 jobs themselves, none of which they really care about, but all of which someone really wants and now is a little bit less likely to get.

I don’t know code as well as I know words, but it looks like the same story is playing out there. SemiAnalysis has started publishing this chart of the number of Claude Code GitHub Commits Over Time. The number, let me tell you, is going up. It’s going up big time.

SemiAnalysis

“4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory, we believe that Claude Code will be 20%+ of all daily commits by the end of 2026. While you blinked, AI consumed all of software development.” Claude Code Is The Inflection Point, is what this means.

And again, I am not a coding expert, but like…

Of course the machines that can happily print tokens at high speed 24/7 are going to produce more code than us meatbags. I can’t imagine what a similar graph of “% of Words Committed to X” would look like. We have to be approaching 50%; we might be way past it. That does not mean the Singularity is Near.

I think if you asked me how I’d position myself on an AI trade, it would be something like: short ASI, long tokens.

As so perfectly captured in Tool Shaped Objects:

The market for feeling productive is orders of magnitude larger than the market for being productive. Most people, most of the time, want to click and watch the number go up. They do not want to be told the number is fake. They will pay— in time, in attention, in actual money— to keep the number going up.

But his is a “come for the tool” view of token demand, and misses the “stay for the network” part. It’s not just that we want to watch our own number go up. It’s that if everyone else’s numbers are going up, we need ours to go up just to keep up.

More words, more applications, more code, MORE, in the world’s stupidest Red Queen’s Race.

I don’t blame the tools. We do this all the time. Venkatesh Rao wrote about Premium Mediocre just two months after Attention is All You Need and years before the transformer’s significance became apparent. Kylie Jenner made her face look a certain way because she could afford it, then everyone started doing it, and it got more affordable, and now Instagram Face has become middle class, a costless sacrifice to the gods of vanity, signifying nothing.

Like, what does the number of Claude Code GitHub Commits signify? What does the number of words written by an LLM signify? What do thousands of applications for every single low-wage job signify?

If they cost nothing, they signify nothing.

I started reading this book recently, The Control Revolution by James R. Beniger, written in 1986 and, from what I’ve read so far, criminally underread today. I haven’t read much of it yet, so I’ll need to save a deeper analysis for a future essay, but Beniger has this idea that is relevant to our conversation.

He believes that modern information technology was built in response to the Industrial Revolution, a direct response necessitated by industrial scale and complexity:

Until the Industrial Revolution, even the largest and most developed economies ran literally at a human pace, with processing speeds enhanced only slightly by draft animals and by wind and water power, and with system control increased correspondingly by modest bureaucratic structures. By far the greatest effect of industrialization, from this perspective, was to speed up a society’s entire material processing system, thereby precipitating what I call a crisis of control, a period in which innovations in information-processing and communication technologies lagged behind those of energy and its application to manufacturing and transportation.

Modern bureaucracy, the telegraph and telephone, mass media, computing, sensors… all of this, Beniger says, fell out of the crisis of control precipitated by the Industrial Revolution. The Industrial Revolution massively accelerated production. We could make things faster than ever. But then we had a distribution crisis: how do we move all this stuff? Railroads and logistics solved that. But then we had a communication crisis: how do all of those far flung hubs coordinate with each other? Enter telegraphs, telephones, train schedules, standardized clocks, and the bureaucracy to manage it all. And a consumption crisis: how do we get people to want and buy all this stuff we’re producing, all over the world? That’s where mass marketing, advertising, and eventually consumer culture come from.

All of which I bring up because it seems that whatever the Crisis of Control was, where we did not have enough information technology to deal with the material abundance we were creating, we have the opposite Crisis now, where the information has outstripped that which it was born to control.

Like going off the gold standard, but for information. We can mindlessly and costlessly print information with no reference to the underlying stuff about which it was meant to inform.

This is the fast takeoff, the runaway scenario: information has reached escape velocity. Claude Code GitHub commits are going stratospheric. But what is the relationship between GitHub Commits and Total Factor Productivity? Between GitHub Commits and actual economic output?

A sacrifice needs to cost something.

The bureaucracy was established on the backs of the hard, provable work of making things. Today, the bureaucracy is the thing, the simulacra of productivity that has become more real than the thing itself, and it lets society make things occasionally, if slowly.

An essay is valuable insofar as it costs the writer something. They may have paid the costs in the years of experience that seasoned the piece that took hours to write, like that Picasso story where he asks for $100,000 for a picture that took him 30 seconds to draw. They may pay the costs in hours and hours, weeks and weeks of researching and writing and editing and re-writing and crumpling and throwing into waste basket overflowing with previous drafts and re-writing again, and again, until it’s good enough.

Even if a machine wrote the same essay, word for word, it wouldn’t have paid the same cost. You can read the costlessness of it, feel it, even as the prose improves. We demand a cost.

This is what King David realized. Fifty shekels wasn’t much money to him, and God, being God, could have divined KD’s sincerity with or without the coin. But a sacrifice has got to cost something.

The bad news is: it seems that we’re in a downward spiral. Systemically, I don’t know how we pull out of this. There will be many, many more Claude Code GitHub Commits and Claude-written words in 2026 than there were in 2025, many more in 2027 than 2026, etc.

Anything can look like the Singularity with a dumb enough y-axis.

The good news is: you don’t need to play this game. You can make yourself pay a cost and you will be rewarded: externally, maybe. Internally, for sure. Eternally, I think so.

There is a school of thought that believes humans were brought into this universe to create. To live and struggle and love and sin and pay all of the costs required to create new things. The reason we are here, this belief, is precisely to experience the limitations and frictions that God cannot. We are here to pay the cost.

There is a joy that comes from conviction, and meaning in doing something only you can do, well. And somehow, that soul shines through. People can feel its presence as strongly as they can feel its absence.

I find that whenever I go comically over the top in the amount of work I do on something, spending a month to write an essay that AI could write ~75% as well in minutes, for example, that sacrifice is rewarded. People want to see that you care, that the work cost you something, that you had to make a choice, in order for them to pay a cost to you.

Pay the shekels.

That’s all for today. If reading this made you want to hire real humans, check out Deel.

We’ll be back in your inbox tomorrow with a Weekly Dose.

Thanks for reading,

Packy

Hi friends 👋,

Happy Friday! It’s quasi-warm in New York City, America won two golds in ice hockey (sorry, Sean, Dan didn’t have anything funny to say but we’re pumped), and the good guys just keep on doing things that make us optimistic.

Let’s get to it.

Subscribe now

(1) Form Gets $1 Billion Google Order for 30 Gigawatt-Hour Battery System

Steve Levine for The Information

The Decade of the Battery is charging ahead, now with American batteries.

Earlier this week, Google announced a deal with Xcel Energy to provide 1.9 GW of wind, solar, and battery power for a planned data center in Pine Island, Minnesota, part of a push for hyperscalers to Bring Your Own Electricity (BYOE). Wind and solar are clean, and cheap when the sun shines and the wind blows, but the sun doesn’t always shine, and the wind doesn’t always blow, so the deal includes $1 billion for a nine-year-old startup Form Energy to provide backup.

Form will be providing two kinds of batteries: standard lithium-ion batteries for instant surges of power, and iron-air batteries for longer-term backup.

Iron-air batteries work through a process that’s like reverse rusting. To discharge, the battery breathes in oxygen from the air. Iron metal at the anode reacts with that oxygen and water-based electrolyte to form iron rust (iron hydroxide/iron oxide). This oxidation reaction releases electrons, which flow through an external circuit as electricity. To charge and store electricity, you apply electricity (say, from solar or wind), and the process reverses. The rust is electrochemically reduced back into metallic iron, and oxygen is released back into the air. The iron is “de-rusted” and ready to discharge again.

Iron-air batteries are low power density and slow response, but extraordinarily cheap on a per-kWh basis, roughly 10% of the cost of lithium-ion. They’re perfect for longer-term storage, up to 100 hours, for multi-day lulls in sun and wind. These lulls are whimsically named “Dunkelflaute” events.

Form will provide a 300 MW iron-air battery system for the project, and the batteries can store 100 hours of power, making it a 30 GW-hr system. It will be the biggest battery system by energy capacity in the world when delivered.

This is good news because we love batteries here at Not Boring, and because it’s a rare win for a western battery company in a category that’s been dominated by China. That also means good US jobs; the batteries for this project will be manufactured at Form Factory 1 in Weirton, West Virginia, on the site of a historic former steel mill.

So yeah, maybe AI is going to take all of our white-collar jobs (kidding, we’re on Team Thompson), but powering all that AI is going to create new ones, too.

(2) Proxima Fusion Signs MoU to Build Stellarator Fusion Power Plant

Proxima Fusion, a German startup out of the Max Planck Institute, signed an MoU with the Free State of Bavaria, RWE, and Max Planck Institute for Plasma Physics (IPP) to put the world’s first commercial stellarator fusion power plant on the grid in Europe.

It will need €2 billion to build a demonstration facility, Alpha, prior to the commercial plant, Stellaris, €400 million of which will come from Bavaria and up to €1.2 billion of which may come from the German government. It’s an expensive way to say “Es tut mir Leid” for shutting down all of the country’s nuclear power in favor of coal, but a very cool way to get German energy back on track.

Of course, it’s just an MoU. The plant still needs to be built, and then there’s the tricky matter of achieving Q>1 and, eventually, Q> whatever it needs to get to to be economically viable, but we love the stellarator.

Stellarators are a type of fusion reactor. In The Fusion Race, we wrote:

Designed by Lyman Spitzer at the Princeton Plasma Physics Laboratory, Stellarators are devices designed to confine hot plasma within magnetic fields in a twisted, torus-shaped configuration to sustain nuclear fusion reactions.

Stellarators, while promising, were difficult to design and build due to their complex magnetic field configurations.

But that was the 1950s! We have better technology today. On Age of Miracles, Julia and I interviewed Proxima CEO Francesco Sciortino, who told us something that’s stuck with us: we made Tokamaks first not because they were best, but because they were the easiest to design and manufacture. Stellarators, with their weird twists, were harder to design and manufacture, but closer to the platonic ideal of a fusion generator. Now, with better software and manufacturing capabilities, we can make reactors closer to the platonic idea.

Proxima has a long way to go before it’s putting electrons on the grid, but we’re happy to see progress towards that goal. Gut gemacht, Germany.

(3) Magical Month for Psilocybin

Psilocybin, the psychoactive compound in magic mushrooms, has been classified as Schedule I drug under the Controlled Substances Act since 1970. The classification means that it has “no accepted medical use and a high potential for abuse.”

That “no accepted medical use” part is coming under intense pressure from reality. Last week, Compass Pathways announced that it had achieved its primary endpoint in a second Phase 3 trial evaluating COMP360 psilocybin for treatment-resistant depression. Two doses of COMP360 25mg demonstrated a highly statistically significant reduction in depression symptoms versus control (p<0.001, -3.8 point MADRS difference). This makes Compass 3-for-3 on trials. The company plans to meet with the FDA to discuss a rolling approval submission between October and December 2026, which would make COMP360 the first “classic” psychedelic cleared in the U.S.

The week prior, a Johns Hopkins pilot study of 20 adults with well-documented post-treatment Lyme disease found that psilocybin, given with psychological support, produced significant and lasting reductions in multi-system symptom burden, including improved mood, fatigue, sleep, pain, and quality of life, that persisted for up to six months. The numbers are striking: general symptom scores dropped roughly 40% and mental/physical quality of life scores rose about 13% from baseline, with benefits maintained through the six-month follow-up. The durability is noteworthy, because a placebo would typically diminish over six months.

The Lyme study is especially interesting because it extends the psilocybin evidence base beyond the depression/anxiety lane into a chronic neuroimmunological condition where there are essentially no accepted treatments. If psilocybin is doing something meaningful for post-treatment Lyme, it starts to suggest mechanisms (anti-inflammatory, neural connectivity remodeling) that go well beyond the “it helps you process emotions in therapy” framing.

From a pharma business perspective, one of the biggest challenges with psilocybin is that it just works. There’s no opportunity to keep patients on (and paying for) meds for the rest of their lives like there is with SSRIs. That’s all the more reason we should be taking psilocybin very seriously. If this same logic applies to other conditions, like Lyme, it could be a major win for humanity.

A miracle drug like GLP-1 whose side effects include euphoria and spiritual experiences instead of muscle loss and upset stomach would be magical indeed.

(4) Stripe’s 2025 Annual Letter

Patrick and John Collison for Stripe

Stripe published its annual letter this week. It’s one of the best pieces of economic writing you’ll read all year, because Stripe has some of the world’s best data on the internet economy, and because those Irishmen can write.

Last week, in Power in the Age of Intelligence, I made the case that tech-native category leaders like Stripe are more valuable because they’re going to eat more of their categories than second-best. Stripe is eating. They shared that they did $1.9 trillion in total payment volume in 2025, up 34% year-over-year, and that roughly 1.6% of global GDP is flowing through the company's pipes.

More interesting is the data they have on companies using Stripe. Their 2025 cohort of new businesses is growing 50% faster than the 2024 cohort. The number of companies hitting $10 million ARR within three months of launch doubled. iOS app releases jumped 60% year-over-year in December. GitHub pushes surged 41% between Q3 2024 and Q3 2025. All of their indicators suggest that the pace is accelerating, and the Collisons’ best guess is that it isn’t an anomaly.

The whole letter is worth a read, but three sections stand out.

First, their framework for “agentic commerce” lays out five levels, from agents that just fill out checkout forms for you all the way up to agents that anticipate what you need and buy it before you ask. They’re honest that we’re hovering between levels 1 and 2, but the comparison to the mid-90s — when HTTP, HTML, and DNS were being hashed out — feels apt.

Second, crypto is happening, even if crypto prices aren’t. Bitcoin is down 50% from its October peak, but stablecoin payments volume doubled to around $400 billion last year, with 60% estimated to be B2B payments. Bridge, the Not Boring Capital portfolio company acquired by Stripe, quadrupled its volume. Stripe is betting on stablecoin payments infrastructure hard with Tempo, a built-for-payments blockchain they’re building with Paradigm. Visa, Nubank, Shopify, and even Klarna (whose CEO was once a self-proclaimed crypto skeptic) are already testing it. Mainnet is launching soon.

Third, the letter closes with what might be its most important idea: we live in a “Republic of Permissions.” Technologies succeed or fail not just on their merits but on whether the web of regulators, committees, and courts lets them through. The Collisons cite Joel Mokyr’s Nobel-winning work on how culture, not just capital or technology, drives progress. They argue that AI could transform drug discovery, nuclear could deliver energy abundance, and drones could slash logistics costs, but only if we don’t let “a slurry of local ordinances harden into a blockade.” Hear hear.

It’s a letter about payments that’s really a letter about whether civilization can keep up with its own tools, co-written by a guy whose website has the canonical list of projects from a time when we were able to build fast. They think we can.

“We’re reminded of the phenomenon of falling into a large black hole… We write this letter at what may well turn out to be the advent of a different and hopefully much more beneficent singularity. While much around us in 2026 feels similar to prior years, it is also clear that the next decade will look very different to those just gone by.”

(5) Vitamin B2 and B3 nutrigenomics reveals a therapy for NAXD disease

Arc Institute in Cell

Pulling one of the most interesting entries from Ulkar Aghayeva’s Scientific Breakthroughs this week above the paywall. Here’s Ulkar on a fresh finding out of Arc Institute:

It may seem that vitamin biology has been figured out decades ago (for reference, all 13 classical vitamins were discovered by 1948). But this paper shows that there’s still a lot of unexplored territory in nutritional genomics.

Instead of starting out with a target disease and trying to find a drug that treats it, they chose the well-known vitamins, B2 and B3, and did a genome-wise CRISPR screen in K562 cancer cells to identify genetic diseases responsive to vitamin supplementation. NAXD, a repair enzyme essential in redox biology, emerged as the top hit for vitamin B3. Mutations in NADX are known to be lethal in early childhood. The team generated knockout mice that showed very similar disease profile to humans, and adding vitamin B3 to their food from birth increased their lifespan more than 40-fold.

How many other diseases are out there that could be so easily cured with various vitamin supplementations?

Additional sources: Arc Institute blogpost; twitter thread

EXTRA DOSE:

• Scientific Breakthroughs from Ulkar

• Jeremy Stern on Shyam Shankar

• Alex Konrad on William Hockey

• Mario joins Hummingbird

Read more

Hi friends 👋,

Happy Friday and welcome back to our 181st Weekly Dose of Optimism!

This week has it all, from transformers to aliens.

We are also introducing a new segment for not boring world members: Scientific Breakthroughs from Ulkar Aghayeva. In December, we shared Frontier of the Year 2025, in which Ulkar, Gavin Leech, and Lauren Gilbert reviewed 202 pieces of scientific news from the year and assigned a score based on a probability each generalizes and how big it would be if they did. We loved it, so we asked Ulkar to do a roundup of the most important stories in science, which she did this week.

It’s our attempt to bring you even closer to the frontier. I hope you enjoy it.

Let’s get to it.

Today’s Weekly Dose is brought to you by… not boring world

The Dose is for everyone, but we’re adding more and more great stuff behind the paywall for not boring world members. We also have a deep slate of co-written essays in flight and going out to members over the next couple months. If you want to support not boring, help us make it as great as it can be, and stay at the cutting edge of technology, science, and business…

Subscribe now

(1) Heron Power Raises $140M to Build 40GW Solid State Transformer Factory

Everything that can go electric economically will, including the grid itself.

Heron Power, founded by former Tesla SVP of Powertrain and Energy Drew Baglino, raised $140 million in a Series B led by a16z American Dynamism to build solid state transformers. This is pure not boring catnip: power electronics from the guy in charge of a lot of the Tesla power electronics stuff we wrote about in The Electric Slide, a16z putting its new funds to work, and fixing the grid by attacking one of its key constraints.

The constraint is that transformers, which take electricity at one voltage and converts it to a different voltage (step up or step down) using coiled wires, are outdated (the fundamental design hasn't really changed since the late 1800s and is based on a one-way grid), passive (electricity goes in, gets stepped up or down, and comes out), massive (built with 10 tons of grain-oriented electrical steel and copper submerged in oil), and worst, in desperately short supply. Lead times stretch up to 24 months, U.S. manufacturing meets less than 20% of demand, manufacturing has been moving to China, prices have spiked 60-80% since 2020, and demand is projected to double in the next decade.

So Heron is building Heron Link, a modular solid-state transformer that uses wide-bandgap semiconductors instead of scarce electrical steel and copper. It’s starting by selling to solar and battery farm operators and data centers, who represent huge and growing demand and for whom the solid state transformer eliminates the need for inverters. Traditional transformers can’t handle direct current, so an inverter turns DC from a solar panel or battery into AC. Heron Link can do that, so customers can skip the inverter. HL is software-defined, meaning it can use software to regulate voltage and frequency to actively manage grid stability. And it’s modular, meaning that if one of the device’s tens of power conversion modules fails, it can be swapped out in ten minutes. They also contain lithium-ion batteries that can discharge quickly to provide 30 seconds of power to smooth the transition to backup power sources.

The funding will go to building a factory that can produce 40GW worth of Heron Links per year. At 5MW per Link, that’s 8,000 transformers.

Heron isn’t the only company making solid state transformers. I hope they all do great. But I do want to call out one competitor: Raleigh-based DG Matrix, whose co-founder and CTO is Subhashish Bhattacharya, a professor at NC State University who is a long-time collaborator of… B. Jayant Baliga, the inventor of the IGBT, The GATEway of India, the hero of the Power Electronics section of The Electric Slide!

Anyway, if we have a lot of smart people using wide bandgap semiconductors to fix the grid, the future of American energy is going to be SiC.

(2) DoW Transports a Valar Atomics Nuclear Reactor to Utah on a C-17

Isaiah Taylor and Valar Atomics

The title pretty much says it all here: the Department of War transported a Valar Atomics Ward nuclear reactor from California to Utah on a C-17. It’s another sign that the government is serious about turning on reactors by the United States 250th birthday on July 4, 2026. Not much more for me to add here other than to watch the video and the others in Valar’s Operation Windlord series.

In other nuclear news, Lockheed Martin is investing in Radiant Nuclear, which we wrote about in 2024, as it prepares to turn on its first reactor at the DOME this summer. Defense is starting to play offense on nuclear.

(3) Single vaccine could protect against all coughs, colds and flus, researchers say

James Gallagher for the BBC h/t Simon Taylor for the find

Researchers at Stanford have built a vaccine that protects against… everything. Like, everything they’ve tested, it works against. The nasal spray works by putting immune cells in the lungs on permanent “amber alert,” ready to fight whatever shows up. In animal tests, it reduced viral breakthrough by 100-to-1,000-fold and worked against flu, Covid, common cold viruses, two species of bacteria, and even house dust mite allergens. Prof Bali Pulendran calls it “a radical departure from the principle by which all vaccines have worked” since Edward Jenner figured out the originals in the 1790s.

It’s early (animal studies, not human trials) but the team is planning deliberate infection studies in people next. The big questions are whether the effect translates to human lungs, how long the protection lasts (about three months in mice), and whether keeping the immune system dialed up causes friendly fire. The researchers don’t think it should be permanent; they envision a seasonal spray at the start of winter or a rapid-deployment tool at the start of a pandemic to buy time while a targeted vaccine is developed.

That pandemic use case alone is worth getting excited about. One of the hardest lessons of early 2020 was how long it took to develop, test, and distribute a vaccine while people died waiting. A universal nasal spray sitting on the shelf, ready to deploy on day one, could change the math on the next pandemic entirely.

And if the seasonal version works, imagine a world where “cold and flu season” just no longer exists. As a parent of little kids, please for the love of God work in humans.

(4) DeepMind Veteran David Silver Raises $1B for Ineffable Intelligence