Replit: Remix the Internet
Replit Raises $80M Series B to Bring the Next Billion Software Creators Online
Welcome to the 1,010 newly Not Boring people who have joined us since Monday! Join 91,940 smart, curious folks by subscribing here:
Hi friends 👋,
I’ve been excited to write today’s piece since I first met Amjad Masad, Replit’s founder, back in July. The more I’ve dug in since then, the giddier I’ve gotten.
The post is a Sponsored Deep Dive x Investment Memo. You can read my Sponsored Deep Dive thought process here, but the TL;DR is: I only write about companies I would invest in, and in this case, Not Boring Capital invested in Replit’s Series B.
I genuinely believe that Replit could become one of the most impactful and important companies in the world. It might be hard to see it today — it looks like a toy — but Replit has the potential to make the internet read-write-remix, spread code literacy, and give a billion people the chance to build businesses online.
Let’s get to it.
Replit: Remix the Internet
// I’ve started and restarted this essay more times than any I’ve ever written. It’s a tough one. It’s hard to capture the full breadth of what Replit is trying to build, and how, and what the world will look like if they succeed. That’s partially because it’s impossible to predict, because Replit is in the primitives business. It’s laying the foundation on top of which one billion people will build the future. After spending hours and hours researching, talking to Replit’s CEO Amjad Masad, and writing this piece, I’m convinced that Replit has the potential to be one of the most valuable and impactful companies in the world. But I probably need to explain why, and to do that, I need to just pick a place to start and get going. So let’s start here.
Everyone’s first line of code has to be written in Replit.
Replit’s mission is to bring the next billion software creators online and accelerate a shift towards a world where the combination of software and the internet is truly The Great Equalizer. The company believes that in the modern world, the ability to write code should be as ubiquitous as the ability to write essays. It wants to lead the transition from a stacked model of software creation to a networked one, and to create the world’s first massively distributed and collaborative operating system.
Replit’s vision is mind-achingly enormous, like the “the world if…” meme brought to life:
But when you’re setting out to build something as big as Replit wants to become, you need to start with a clear and ambitious proximate objective, something to tackle right now.
That’s why Replit wants to be the first place that anyone writes a line of code. Then it can become the first place they build an app, the first place they collaborate on software, the first place they make money online, and ultimately, the place that they turn whenever they want to create software.
It’s a straightforward goal with nuanced implications.
For example, it means not targeting experienced developers to start, even though they’re the more valuable users today. If they’ve already written their first line of code elsewhere, they definitionally can’t write their first line of code in Replit.
It also means that Replit needs to make it really easy to get started by putting everything you need to code something simple in one place, accessible from anywhere in the world with an internet connection whether you’re on a brand new MacBook Pro, a Chromebook, or even an Android phone.
And it means making it multiplayer from the beginning, to create a modern development experience that includes collaboration, like people are used to with products like Google Docs and Figma. A networked operating system demands networked development.
The approach comes with trade-offs. As it stands today, a large company couldn’t build most of its software in Replit. It’s more limited than a local setup including best-in-class point-solutions like Microsoft’s VS Code, Github, Heroku, Docker, AWS, npm, and postgres. It’s slower. It’s not industrial-grade. But that’s not the point.
Replit wants to build the future, and to build the future, you don’t rebuild the old thing for the old customers. You build something new that starts out looking like a toy, and you build for the underserved, not just for moral reasons, but for strategic ones. It’s disruptive innovation meets the compounding power of young users. It’s more Roblox than Github. As Replit’s CEO and co-founder Amjad Masad explained it to me:
To discover the future, you need a population that will grow up with the technology and partner with you in figuring out where software is heading, how we collaborate, and what the implications of things like web3 and AI will be.
Partnering and growing with underserved (here: young) customers requires an uncommon patience and focus on the mission. Replit is already a decade old if you count its earliest instantiation as repl.it, an open source project that Amjad built growing up in Jordan. That project led him to Codecademy and Facebook before returning to the original mission and rebooting Replit as a company in 2016. Today, Replit, Inc. is in year five of a multi-decade plan.
That plan starts with becoming the first place that everyone writes code, and growing with them symbiotically as they grow. That means giving new software creators everything they need to start creating, accessible from anywhere to anyone with a laptop, tablet, or phone.
It ends with Replit becoming the world’s first massively distributed and collaborative operating system, an open and extensible platform that anyone can build on top of and plug into to improve the platform itself, and make money by doing so. The more people who know how to code, the more people there are to improve the platform, the richer the ecosystem becomes. It’s a strongly positive sum network effect.
When Replit is successful, everyone will be able to remix the internet.
Because of the emergent nature of composability, it’s impossible to predict what Replit will look like in a decade, but it’s not impossible to predict that the results are going to be special. If the company gets everything just right, I think that it has the chance to be a $100 billion+ company that reshapes the way that software is developed and by whom, and by extension, who can build the internet.
It’s a multi-decade vision, and building towards it is going to take a lot of time, money, and smart people. That’s why today, Replit is announcing an $80M Series B led by Coatue at an $800M valuation.
Actually, I’m just going to call it: Replit will be a $100 billion+ company, but more importantly, its users will generate multiples of that number in value for themselves.
Not Boring Capital invested in this round and I couldn’t be more pumped. Replit embodies so many of the themes and ideas that we talk about in Not Boring -- composability, decentralization, disruptive innovation, the compounding power of young users, web3, and more -- and it’s building a future I want to see exist.
To understand the company’s enormous potential requires projecting the course of each of those themes forward and where they might interact, and then weaving together those different strands into one coherent narrative. Honestly, it’s been one of the hardest pieces for me to write because each thread I pull opens up new paths for exploration, a new “most exciting thing” to elucidate.
So maybe we fast-forward to the end for a second.
Imagine this: a world in which anyone, anywhere can learn to code, snap together composable services on top of a shared data and compute layer to build products, and monetize those products instantly and continuously, all in the same place. Where everyone can manipulate code like we manipulate words, creating as much richness and novelty in the programs we use as in the books and essays we read.
We’re going to go down a winding path to get to that world, and I hope I can do it justice. Join me as we explore:
The Internet Wants to Be Open
From Jordan to Replit
What is Replit?
Hockey Sticks and Dollar Signs
Compounding Power of Young Users
Building the Operating System for the 21st Century
Before we get to Replit’s story, we need to talk about what’s at stake.
The Internet Wants to be Open
It shouldn’t be surprising that the internet is run by large, centralized companies today. Honestly, our fault.
The early internet, from the 1980s to the early 2000s, was largely open and decentralized. It was built on top of a series of open protocols that anyone could build directly on top of, like HTTP for websites, SMTP for email, SMS for messaging, IRC for chat, and FTP for file transfer. The benefit was that these protocols were generally agreed upon and not subject to change; I could build a website on HTTP and if people had my website address, they could go directly to my site, not intermediated by anyone else.
But there were some major challenges, which I covered in The Value Chain of the Open Metaverse:
Stateless. Web 1.0 protocols were stateless, meaning that they didn’t capture state, or user data. “Capturing user data” has negative connotations today, but stateless protocols meant that website owners didn’t even know whether I’d visited a site before, and couldn’t tailor experiences accordingly.
Missing Protocols. Web 1.0 didn’t have standard protocols for many of the things that power the internet today: payments, search, apps, social media, commerce, credit, and more.
Protocols Didn’t Make Money! Imagine developing HTTP, seeing trillions of dollars worth of value being built on top of it, and not being able to participate in the upside aside from some speaking fees, consulting gigs, and book sales. Oof.
Too Technical. You needed to be technical to build a presence on Web 1.0, which meant that regular people who didn’t know how to code were left out.
Web 2.0 (mid-2000s to present) emerged as entrepreneurs recognized the holes in Web 1.0 and built products to fill them in, and capture the value in the process. These companies didn’t just capture state, they aggregated it, building up huge troves of valuable user data. They made it so that anyone could participate and build a presence -- think about how easy it is to set up a Facebook page versus coding a website. They wrapped existing protocols in frictionless user interfaces and created de facto products where no protocols existed.
We’ve covered some of the challenges with Web 2.0 in Not Boring, along with some of the ways that web3 might help solve them. In many ways, web3 is Web 1.0 with most of the holes filled:
Shared State. Blockchains are big databases that capture and retain shared state.
New Protocols. Each new lego is a new protocol that others can snap into.
Protocols Can Make Money. From Bitcoin to Ethereum to Uniswap and beyond, successful web3 protocols can be worth hundreds of billions or even trillions of dollars.
But one thing that still hasn’t been solved is that most of us aren’t technical enough to actually create software. Fewer than 30 million people in the world know how to code. Anyone can now own the protocols and apps they use, but most of us (certainly me included) still can’t build them, or build on top of them.
A truly open, decentralized internet will be one that a billion people can build on and remix and compose, where anyone can interact directly with protocols, APIs, and services, and build protocols, APIs, and services themselves. It will be a world in which people can design their own internet experience, and design software that can succeed not by being 10x better than the existing solutions or 10x bigger than competitors, but by better serving the millions of small niches that make up the internet.
The commonly way to describe the evolution of the internet is as follows:
Web 1.0 was read-only
Web 2.0 was read-write
Web3 is read-write-own
Ownership is a critical piece, but it kinda breaks the read-write metaphor. The thing that prevents centralization ultimately will be how many people are able to create unmediated homes on the internet directly on top of protocols and shared state. People should be able to fork and build and tinker and tweak to build their own experiences on top of global, decentralized databases. When that happens…
Web3 will be read-write-remix.
When I wrote that on Monday, I had Replit in mind. We’ll talk more about that world in a bit, but I wanted to set the stage with what’s at stake.
For now, let’s go back to the beginning of the story, when Replit was born of a pressing need felt by a young Jordanian hacker.
From Jordan to Replit
It’s possible that only Amjad Masad could have built Replit. His combination of technical skills and lived experience might be exactly what’s required to reimagine who gets to create and monetize software, and how, in a world in which talent is equally distributed but opportunity is not, yet.
Amjad’s story proves two things:
Even ideas that seem inevitable in hindsight are really hard and uncertain in real-time.
Surviving the roller coaster is easier if you’re focused on a personally important mission.
Amjad Masad was born in Amman, Jordan. When he was very young, his dad, who was working two full-time jobs, running his own business after coming home from his 9-5 job, bought the family its first computer. Back then, even getting a computer running required a little bit of basic (pun intended) programming, guided by a thick instruction manual, and young Amjad learned as his dad did.
Amjad’s childhood looked a lot like Replit’s product roadmap. By age six, he could do some light programming. By age 12, he was programming in Visual Basic. By age 15, he sold his first piece of software, a LAN gaming center management software, and started a small business to maintain the computers and games at the stores that bought it.
In high school, he walked a familiar hacker’s path: originally guided by a teacher who gave him advanced lessons in math and computer science on the side, Amjad became bored and unruly when the teacher left the school and the lessons stopped. His parents made frequent trips to the principal’s office to bail him out. But he was smart, and he escaped high school.
For college, Amjad went to Princess Sumaya University for Technology (PSUT), where he once again found himself understimulated and underperforming.
According to a 2013 profile, he was dropped from the same calculus class twice, not because he wasn’t good at math, but because he never showed up. Instead, he was spending his time coding for startups and bigger companies, and hacking on the earliest version of what would become repl.it. But he did want to graduate, so to prove his merit, he did what anyone would do: he hacked into the university’s network and exposed vulnerabilities to the administration. Instead of kicking him out, they offered him a deal: work with the IT team to rebuild the network’s architecture and security, stay in school.
Finally, after six tedious and soul-crushing years, he graduated in 2010 and went to work for Yahoo!, which, you guessed it, was tedious and soul-crushing work to a creative hacker like Amjad. This time, to keep his brain occupied, he started trying to learn a bunch of different programming languages. Only problem was, Amjad didn’t own a computer, so after work, he would go to internet cafes, where each night, he had to set up his programming environment from scratch, all while he was paying for computer time. Worse, he couldn’t save his progress, so each time, he had to restart, which slowed down his progress even further.
He told Japanese blog gihyo:
My goal was to not only learn, but be proficient in four programming languages in a year. You can pick up the basics of a language really fast, but to really learn it, you need to practice a lot. But the lack of a good, portable IDE that is accessible from anywhere and anything with a browser makes it too hard to start.
Google Docs had recently come out, making it easy to write and collaborate in the browser (yes, there was a time in recent history when that wasn’t a thing), and Amjad wanted something similar for code. It didn’t exist, so once again, he did the most hacker thing he could: he built a solution.
He teamed up with classmate Max Shawabkeh and future wife/co-founder Haya Odeh to build a program that could support cloud-based programming environments from a web browser. He didn’t need a full-fledged integrated development environment (IDE), just something to play around with and learn on, like the interactive REPLs he’d learned about studying SCHEME, a LISP dialect taught to first-year MIT computer science students.
So they built repl.it, and open sourced the project.
Meanwhile, across the ocean in New York City, two young Columbia grads, Zach Sims and Ryan Bubinski, discovered the project on Github and realized that the interactive, web-based programming environment Amjad built would be perfect for the coding education startup they were building: Codecademy.
Within two days of launch in August 2011, Codecademy signed up 200,000 users, which meant that 200,000 people around the world were suddenly using the code that Amjad had built in internet cafes in Jordan. A month later, Zach and Ryan emailed Amjad asking him to collaborate. Amjad was already planning to quit Yahoo! to build his own business on top of repl.it, but the idea of joining forces was appealing:
In talking to Ryan, I realized we had the same vision of what we want to do, of how we want to make the programming accessible to everybody. We had similar thoughts about coding and what we call “the new literacy” and how everybody should learn how to program.
The trio worked on projects for a month, and in November, Zach flew to Jordan to convince Amjad to move to New York and work with them as a founding engineer at Codecademy. They went back and forth, debating equity, and as they drove Zach back to the airport, Amjad told him that he wouldn’t make the leap at the lower equity package Zach was offering. So in the taxi, Zach upped the offer, Amjad signed, and in January 2012, he moved to New York.
After two years building Codecademy, during which the company raised a $10 million Series B from investors including Index, Union Square Ventures, Richard Branson, and Alexis Ohanian, Amjad decided that if he was going to live halfway around the world, he wanted to live in the home of hacker culture: Silicon Valley.
He left Codecademy and headed west before figuring out his next stop, and quickly landed a job at Facebook. He was attracted to the company by its strong engineering culture, but was specifically drawn to internet.org, the company’s project to bring internet access to less developed countries. It was another clear step in Amjad’s personal quest to bring the power of software to more people around the world. But the way Facebook works, you get hired by the company, and then interview with teams inside of the company. And the internet.org team didn’t have room for Amjad.
But at the same time, a small team at Facebook was working on building a new open source framework that would make it easier for engineers to create software for mobile: React Native. Amjad joined that team as a founding engineer, and spent nearly three years building it out.
All the while, while Amjad was working towards his personal mission of making coding more accessible at Codecademy and Facebook, repl.it continued to grow in the background. Still, no one had built the product that he had thought about building back in 2011 on top of it. As he wrote in 2018:
I'd been waiting for someone to start this startup for years -- a website where people can not only learn how to code but can also build, collaborate with others, and ship software.
Amjad wanted that company to exist, but he wasn’t sure he wanted to be an entrepreneur. This recent thread captures all of the things he did to try to not start Replit as a startup.
Facebook wouldn’t let him build it as an internal project. Merging with existing startups didn’t work. Y Combinator rejected Replit three times (before ultimately accepting them, which you can read about here). He attempted to bring in an outside CEO but couldn’t find one who believed in the vision as strongly as he did.
So Amjad did some soul searching. Specifically, he dug into Danish philosopher Søren Kierkegaard’s work on despair, heavy stuff that deals with peoples’ internal tension between the finite and the infinite. (My man contains multitudes.) You should read Amjad’s thoughts in this April 2016 blog post for more color, but ultimately, he realized that entrepreneurship was a means of reconciling the finite (just doing) and the infinite (dreaming and imagining).
I’m about to drop a long block quote from that essay, which Dan always advises me not to do, but I think it’s important because it both directly influenced Amjad’s decision to start Replit and indirectly shaped the company’s product and strategy:
I view entrepreneurship as means of reconciling the infinite and finite. You venture into your imagination, gather knowledge, and dream about a better world. But you have to bring some of that back to earth. You take a step -- no matter how small -- towards your imagined world in the real world.
This is not a one time thing, it's a recursive process. If your imagined possibilities became real then that's your new "finite". The process restarts and that's how we make progress.
I see this as the perfect framework for the work we do in technology. There is always a tension between what is and what could be.
Taking a step into the infinite could be hard, but the hardest thing of all is bringing something back to the finite. The real art is coming up with the smallest possible task that can be done here and now.
Amjad wrote that post on April 14th, 2016. That same month, he quit his job at Facebook, resolved his Kierkegardian tension, and founded Replit.
(Actually, he launched “Neoreason” to build a suite of products that augment human reason, starting with programming, but quickly adopted the name Replit from the open source project when they realized how hard it would be to build just that one product really well.)
The early days of Replit were a family affair. Amjad started the company with his wife, Haya Odeh, a talented designer, and his younger brother, Faris Masad, a strong coder in his own right. Amjad had trained Faris in math from a young age by writing small games to teach him addition, subtraction, multiplication, and division.
The company the trio launched is *rips bong* literally built to help people take a step – no matter how small – towards their imagined world in the real world, and then to iterate recursively towards mastery. 🤯
So what is Replit?
What is Replit?
Replit is the easiest place to start coding, build, deploy, and host apps, collaboratively.
One way to think of it is like Figma for coding. It’s everything someone needs to code, all within the browser and with real-time collaboration built-in, that’s simple enough for beginners to pick up in minutes (I make all my beautiful designs in Figma).
Another way to think of it is like Roblox for software. Just as teenagers can build, host, and sell games and items all within Roblox, teens can build, host, and sell software in Replit. Unlike Roblox, though, Replit is plugged into the entire internet.
The way Amjad has described it in the past is Google Docs for coding. Instead of downloading software locally and sending files back and forth a la Word, Google Docs made it possible to write in the browser and collaborate in real-time. Replit’s pushing the same transition in a much more complicated space.
Analogies only go so far, though. Replit is building something novel and sprawling from the ground up, pushing code to do things it didn’t know it could, in order to build the ideal environment to start creating and selling software.
What does that environment look like?
The company’s name is a nod to the REPL -- the Read-Eval-Print-Loop interactive coding environment -- but that’s just one piece of what it offers. The description that shows up on Google is:
Replit is a simple yet powerful online IDE, Editor, Compiler, Interpreter, and REPL. Code, compile, run, and host in 50+ programming languages.
Replit is more than a REPL and more than an IDE. It’s the first software to fully integrate the entire computing environment on top of simple yet scalable abstractions and make it multiplayer from the start.
Amjad expanded on what Replit gives users in Internet of Fun: “Every repl on Replit is a full computer that comes with access to the network, a database, storage, and access to its code, user profiles, and other primitives.”
To appreciate what that means, let’s take a step back and look at how it works today. In its 2020 Series A Deck, Replit included a slide that showed all of the tools that developers needed to stitch together to build and deploy an app:
Coders need to set up a code editor or IDE like Microsoft Visual Code Studio, host source code in Github (also owned by Microsoft, find and connect to APIs with RapidAPI, host the site on Heroku, control log data with logdna, and when they get stuck, turn to stack overflow.
That’s six different pieces of software, some local and some web-based, to download, set up, create accounts for, pay for, and jump back and forth between, and that’s just the basics. The company’s 2021 Series B deck included a few more logos:
If that was a lot to read, if I’ve lost you… that’s the point!
It’s a lot to even read about all of the different tools that it takes to create, deploy, and host a software product. Actually setting up all of these tools, figuring out how they work, and jumping between all of them would take a new software creator days or weeks. Back in 2012, I taught myself to “code” to build a simple website, and the “coding” part was really fun, but even figuring out which tools to use and then setting it all up was a nightmare. It made me stop “coding.” (Well, that and the fact that I wasn’t any good.)
In a move that would make Jim Barksdale proud, Replit simplifies the experience of beginning to code by bundling all of the things you need to create and sell software. This table from its homepage is a quick overview of many of the things included with a Replit account.
Replit is a place to code and collaborate in practically any language, instantly.
It supports nearly every programming language, and is beginning to open up to let creators submit their own.
If your code relies on a particular package, Replit pulls it in automatically.
If you want to run code from Github, you can plug it into Replit and run it immediately, with all of the things it needs to work.
If you get stuck, you can turn to the community of over 3 million active coders for help.
If you need something that Replit doesn’t offer, you can turn to its marketplace of plugins and APIs.
Once you’ve built a product, you can deploy and host it at the click of a button.
If you want to sell your product, you can do it right within Replit.
It’s everything software creators need to build simple games, apps, bots, and products, all in one place.
Today, none of those components is as powerful as its best-in-class counterpart. That’s fine. New users -- 80% of Replit’s users are outside of the US and 50% are under 18 -- are just looking for a place to hack around and learn and build their first product. That means simplicity first, then power.
The heart of the simplicity is the in-browser IDE, which lets users spin up new repls, or coding environments, in a second. Here, I’ll spin one up right now.
Boom. In the 15 seconds the gif recorder gave me, I had a node.js IDE up and running. I could invite teammates to code with me in real-time, publish our work, host it, put it on the world wide web, and start making money just as easily.
Unfortunately, I still don’t know how to code. While Replit sold to schools and coding education programs, it’s not a coding school itself. Instead, Amjad recommends taking some intro courses or watching videos online through programs like Free Code Camp and then jumping in to start hacking and iterating your way to knowledge.
Once you have the basics down, you can dive back into Replit and start coding.
You can start from example templates, or fork any other user’s project to get their full code and learn while you remix their work.
You can run any changes you make right in the browser to see how your tweaks impact the product and get a better feel for how the code works.
Importantly, the ability to fork and run code all in the same place turns code from a static thing, like it is in Github, to something that’s alive, always ready to be remixed and run.
You can do everything I just wrote about in single-player mode, hacking away alone in your basement like a caricature, or in multiplayer mode with collaborators from around the world. I’d try to describe how multiplayer works, but this video does it better than I could.
Finally, when you and your friends have built your product, Replit automatically hosts it on its servers for free. You can share the fruits of your labor with the world as easily as sharing a link. If it’s really good, it will show up in the Apps gallery, where millions of Replit users can find, run, and remix your work.
Figuring out how to even start coding is intimidating. With Replit, you still need to learn how to actually code, AI won’t do it for you (yet…although this Replit x OpenAI Codex drop uses AI to explain how any code you highlight works), but everything else is taken care of.
All told, Replit has built a product optimized for getting as many people to start coding as possible, and to make sure that they do it in Replit.
Is it working?
Hockey Sticks and Dollar Signs
When we left our heroes, Amjad, Haya, and Faris had decided that if they wanted Replit to exist in the world, they would have to build it themselves. And they were going to have to fund it themselves, too.
Haya and Amjad agreed to give themselves $20k to get the company off the ground until they could find a revenue model or get the company off the ground. While they sold their code evaluation infrastructure as an API to earn some money, the free (and expensive to maintain) product was growing a lot faster.
Luckily, in October 2016, Roy Bahat at Bloomberg Beta took a chance on Replit, leading a $600k round at a $6 million valuation that they called a seed, but was really like a pre-seed. Around that time, they were rejected by Y Combinator for the first time.
Having secured a little breathing room but not enough to get comfortable, they agreed that they needed to be “default alive,” a Paul Graham concept that meant earning enough revenue to keep itself going even if outside funding dried up. They decided to start selling a paid version to schools, which they called Classrooms and recently rebranded to Teams for Education. As Amjad has said on podcasts, that meant operating in two markets in which it’s traditionally been tough to build big businesses: Developer Tools and Education.
Despite the challenge, after a few months, Replit was on a $150k revenue run rate. It was a good start, but the team realized even then that it wasn’t a scalable model. Meanwhile, though, the free REPL product kept growing and growing on its own and was starting to generate buzz on Hacker News, Y Combinator’s popular message board for hackers. Y Combinator had now rejected Replit three times, but its founder, Paul Graham, stumbled upon the product on Hacker News and told YC President Sam Altman to reach out.
On the eve of the application deadline for the Winter 2018 batch, Altman encouraged Amjad to submit a late application, which he and Haya completed hastily that night, including a joke link to a notorious Rick Astley song in lieu of a video submission. The next morning, Amjad and Haya took an Uber to YC in Mountain View and waited outside.
Three hours later, they got the call: they had secured an interview, and rushed inside, where partner Michael Siebel greeted them, deadpan, with their first question: “Why did you Rick Roll us?” Somehow, even though the joke bombed, they got accepted. They raised the standard $120k for 7% from Y Combinator, kind of an official post-seed pre-seed, and entered the program.
One of the things that probably tipped the scale in their favor was that, already, they had about 750k registered users by the time they started at YC.
By the time that a16z led a proper $4.5 million Seed round in October 2018, Replit had blown past a million registered users and were quickly approaching 1.5 million, double the number when they entered YC nine months earlier. The round brought both March Andreessen and Paul Graham, two internet pioneers with a deep understanding of the power of young hackers, onto Replit’s side. Just as importantly, the company announced that it was hosting 250k apps and websites just seven months after launching hosting in March 2018.
For the next two years, the company continued to build new features, new ways for young people to start coding, and they continued to grow. In early 2020, Replit closed a $20 million Series A led by A. Capital. They decided to hold off on announcing the round due to COVID.
By December 2020, it crossed the 5 million registered user mark, prompting this tweet from PG:
By February 2021, when Replit announced its Series A, it had already grown to 6 million registered users, up 122% YoY.
That kind of growth, obviously, attracted venture capitalists. Throughout the spring and early summer, VCs kept reaching out to preempt a Series B. Amjad kept telling them no. Finally, a “too big to ignore” fund came in and kicked off the process. They moved too slowly to seal the deal, but other funds swooped in.
Coatue’s Matt Mazzeo wrote the first term sheet, then another fund came in at 2x the price, Coatue came back a little higher, and another firm came in above that. Amjad chose Coatue, despite the lower price, because he was excited about what Mazzeo and the firm could do to help the company grow and build out its business. When the dust settled, Replit had raised an $80 million Series B led by Coatue at an $800 million valuation and Mazzeo joined the board.
I participated in this round with Not Boring Capital, and among everything else that I’ve talked about, there were three main reasons.
First, Ankur Nagpal, who introduced me to Amjad, was investing, and for Ankur to invest in a round at an $800 million valuation, there needed to be something really special happening.
Second, Replit’s growth was unbelievable. Check out this VC catnip:
Replit had reached 8 million registered users, up from 6 million in February.
Those users had created and hosted more than 13 million websites and apps on Replit in just three years.
Those apps were getting over 10 billion monthly visits.
This slide from the deck summarizes all that growth in one place:
Today, just three months after the round closed in August, Replit has already grown to 10 million registered users who have created 20 million apps and websites hosted on Replit. At any given point, Replit is running half a million concurrent containers. The scale is mind-boggling.
To be fair, though, Replit is still small compared to the big players in the space. Github had 56 million users as of September 2020. Cloud giant AWS accounts for 33% of the cloud computing market. And while Replit didn’t want to disclose revenue, it’s $7/mo Hacker Plan, from which the company currently derives most of its revenue, isn’t likely going to be the main source of revenue at scale. And $800 million is a big price tag.
To overtake strong incumbent competitors over the next decade, Replit is attacking a three-part master plan:
Become the default entry-to-coding environment and grow the worldwide developer population
Become the best place to create and monetize software
Build the operating system for the 21st century
The whole strategy is centered around the third reason I’m so excited, one that deserves a section of its own: the Compounding Power of Young Users.
Compounding Power of Young Users
Whatever platform the best 15 year old hackers are using today, everyone will use in 10 years. That was true of microcomputers in 1980, and of web apps in 2000. As long as teenage hackers love Replit, it’s the future of computing.
– One of Replit’s Investors
One of the essays that I link to most often in Not Boring is Chris Dixon’s The Next Big Thing Will Start Out Looking Like a Toy. “Disruptive technologies are dismissed as toys,” he wrote, “because when they are first launched they ‘undershoot’ user needs.” One way that these toys can become the next big thing is that they improve faster than users’ needs increase.
Another way, which Chris didn’t write about, is that they can find an audience of people who love toys: kids. As those kids’ needs grow, the toy can grow with them, until both are all grown up and in charge.
This is an idea that I call the Compounding Power of Young Users:
If you can acquire the youngest users, retain them as they get older, and continue to attract the new cohorts of young users, you will win over time.
This is one of my favorite strategies because it requires such patience and long-term vision to pull off. It’s why Snap made a comeback last year despite not moving upmarket to capture older users, how Stripe was able to grow by focusing on overserved startup engineers and maturing with them as they matured into leaders. And it’s why I thought that Google should acquire Slack, so that they could combine their younger workplace user bases to take down Microsoft. And it’s why I’m particularly excited about Replit.
Replit has the opportunity to become the best example of the Compounding Power of Young Users in history.
The company’s express goal is to be the place where everyone writes their first line of code. As we’ve discussed, that means intentionally ignoring the upmarket opportunity for now. Many of the products that Replit’s users create are digital toys: games are the most popular projects on the platform.
There was a slide in Replit’s Series A deck that was meant to scare VCs away…
…but should have made any investor with knowledge of the history of technology salivate.
Replit’s young, international users aren’t the most valuable today. 50% of them are under 18 years old. In May, Amjad tweeted an example of one 12-year-old kid who had been coding in Replit since he was 8 years old.
He’s also shared examples of a 13-year-old who built a large crypto mining operation on Replit and a 16-year-old Indian kid who earned his first $1,000 thanks to Replit.
Today, those kids are happy hacking around and learning in Replit, and Replit needs to get as many of them to start coding in Replit as possible while they’re young. That means partnering with schools, even though education is a hard sell. It means making it easier for kids to program what they want to program. They recently launched kaboom.js, a simple game programming library and environment, because they knew that a lot of kids started writing code in Scratch, MIT’s gaming language.
Over time, as those kids evolve and get more talented and build more complex things, Replit’s job is to keep up, or better yet, stay a step ahead. Some of that will come from better software and primitives that the company builds itself, like the Nix-based OS infrastructure it built to support every programming language, but a lot of it will come from the things that the community builds itself.
The more extensions, add-ons, code, and services that live inside of Replit, the less of a reason young hackers have to try anything else. If young hackers build their businesses, and their customer bases inside Replit, the higher their switching costs. Plus, as they make friends and collaborators in Replit, and build their social/professional network in the community, the less likely they are to leave.
If Replit gets it right, they might be able to leapfrog the IDE. In a post of the same name, Amjad wrote that the company realized that some more advanced developers weren’t just using Replit to learn and hack around; they were using it as their primary programming environment:
This seems surprising at first: other than for learning, why would anyone use a more limited tool over a more mature and flexible one?
It turns out this was the wrong question to ask. A better one would be: "If someone had already learned to code on an online REPL and it serves their needs then why would they want to invest in a local setup?"
That’s the Compounding Power of Young Users at work. Now that Replit has wedged its way into its young users’ workflows and habits, those customers are Replit’s to lose as they mature. The burden of acquisition it on Replit’s upmarket competitors.
Over the next few years, Replit will need to squeeze more power into its browser-based environment, to satisfy its users as they improve, while maintaining the simplicity that will continue to feed the top of the funnel with new users. Dixon has thoughts here, too, writing that the toys that become big things ride powerful “external forces: microchips getting cheaper, bandwidth becoming ubiquitous, mobile devices getting smarter, etc.”
To transition from Step 1 – become the default entry-to-coding environment - to Step 2 – become the best place to create and monetize software – Replit’s small but mighty team of generalist hackers, alongside its community, will need to take advantage of macro improvements in technology to pack more punch in the browser.
They’ll also need to give their users the tools to build better ways to monetize their work, and to build a new economy.
In September, Amjad dropped this double-entweetdre:
Three weeks later, Replit dropped this bomb on the internet:
In the blog post announcing Solidity on Replit, the company wrote that “Replit and Ethereum have a surprising amount in common,” so much so that Amjad’s tweet could apply to both.
On the surface, by making it easier to learn and build with Solidity, solo or collaboratively, Replit is running its usual playbook on a new language. But the alignment between Replit and web3 goes deeper than that.
When I asked Amjad how deep down the rabbit hole the company was going to go, he told me:
We’re aligned on the idea that people should own what they create. We’re aligned on the idea of software becoming composable and alive. The most exciting thing about the blockchain is that it makes software live, whereas open source code is dead. Where the world is headed, there’s a coordination problem in open source, but protocols that capture more value make it easier for complete strangers to coordinate.
In the near-term, the company is studying all of the chains, and it wants to support all of the languages. They’re working on a big Rust course given that Solana is a huge driver of adoption, and hosting a hackathon with NFT marketplace OpenSea.
Replit is going deep, soooo…. wen $REPL?
Replit isn’t going to decentralize and launch a $REPL token any time soon. Amjad cited the fact that tokens often end up subsuming projects and overshadowing the important technical things many projects were originally conceived of to achieve.
That said, Replit will introduce value into the Replit network, potentially as a centralized native Replit currency to start. Replit wants to reduce the time it takes someone to go from writing their first line of code to making their first dollar on the platform, and with such a large base of international and under-18 users, Replit currency might be the easiest way to make that happen. In the beginning, that might look more like Robux than $REPL, but over time, the company is excited about the idea of introducing programmable money.
“Building a unit of account into the system will unlock a lot of interesting things,” Amjad told me:
If we give people money as a platform primitive, people will do surprising things. The way I see us is as a toymaker: when you make a toy, you don’t know how kids are going to play with it, you just need to watch and see what they come up with.
Maybe the next big things will start out looking like a toymaker. If it’s successful in its ambitious vision, Replit will help bring about a fully-formed web3, one that’s read-write-own-remix.
Like web3 itself, Replit will be a vortex for talent, creativity, and economic activity online. It might become the operating system for the 21st Century, the platform on top of which much of the internet is built, remixed and monetized.
Building the Operating System for the 21st Century
I’ve been fully glowing in my analysis up until now. As you can tell, I’ve been Replit-pilled. That doesn’t mean that success is guaranteed by any stretch of the imagination; it just means that the risks are obvious enough as to not be particularly interesting.
First and foremost, Replit is surrounded by competitors on all sides.
On the high-end, Microsoft and Amazon loom large. Microsoft is the 800 lb gorilla in programming, with GitHub, Visual Studio, and Azure, and Amazon is both the largest cloud provider with AWS and owns one of the most popular cloud-based IDEs in Cloud9. It’s easy to say that Replit can retain users as they grow, but as it stands today, many will need to move into more powerful environments if they go work for a big company or want to build scale products.
It’s hard not to appreciate the coding and gaming war chest that Microsoft has built up through acquisition. Until writing this piece, I didn’t realize that Microsoft had the most popular IDE (Visual Studio) and code repository (GitHub), plus Minecraft and XBox and a host of other products, all on the side of its desk. Plus, it’s transitioned Office to the cloud well under Satya Nadella after a slow start. It won’t go down easily. I would be surprised if Microsoft doesn’t offer Replit a whole lot of money to stave off the threat one day, and I would be more surprised if Replit sells; it may or may not have turned down some pretty big offers in the past.
And then there’s the no-code movement. Maybe being able to string together no-code tools will replace the need to code for most people (although that starts to look like coding itself).
But competition is a known quantity, and one that Amjad told me he’s not particularly worried about. As long as Replit keeps growing and retaining young coders like they have been, the future is theirs.
Second, there’s business model risk. Today, Replit makes the majority of its money via subscriptions, but that’s not how it plans to monetize in the future. It has the outlines of its future business model sketched out, but its ability to make the switch is yet to be proven.
The third big risk is that transitioning from a Replit-controlled coding environment to a full fledged platform and economy is a really hard transition. If the underlying technology doesn’t get good enough to match local solutions in the browser, or if those local solutions improve at a faster rate than Replit can, then the dream of growing with young users could be dead on arrival.
Then there’s the fact that, depending on whose definition you use, Replit isn’t even a platform yet, let alone the operating system for the 21st Century.
That’s a crock of shit. This isn’t a platform. A platform is when the economic value of everybody that uses it, exceeds the value of the company that creates it. Then it’s a platform.
Amjad has a slightly different definition: a platform is when most of the functionality is built by developers not employed by the company.
Today, Replit may be a platform based on the Bill Gates definition because most of its users use Replit for free, and many are making thousands of dollars or more from what they build, but by Amjad’s definition, it’s not yet a platform. Most of the functionality is still built by Replit’s small but mighty team of hackers.
But that definition, along with everything else we’ve covered up to now, hint at what Replit is building towards.
Replit envisions a world in which a billion people learn to code and build products on Replit. Many of those products will be built for the broader internet, but many will also be built to make Replit itself more powerful.
It’s a double-compounding power of young hackers: as they develop, they’ll get good enough to improve the platform itself.
By rebuilding its infrastructure on Nix, Replit just tapped into an infinite network of open source code. Next, Replit might let an infinite network of builders tap into it by opening its APIs, open sourcing a big part of the platform, and letting people plug in and build their own extensions. Replit is really a primitives company, and composability will be a key piece of its strategy. That could take a variety of forms:
Users will build tools and services for other users, and using programmable money as a primitive, will get paid whenever their services are used.
It could build its own hosting infrastructure, or tap into a global network of smaller cloud hosting providers to compete with AWS in a more decentralized way.
Replit could let users charge to let other users fork their code, with royalties streaming back to the original creator every time the remixed code is used commercially.
Replit might host a software, API, and infrastructure marketplace where its users can find the legos they need to build their things.
More likely, just as Replit automatically pulls in the right packages today, it could automatically pull in the right piece of software at the right time for its users, and make payments happen automatically. Users might even be able to add mini-paywalls with annotations in the code base.
In this world, Replit would make it as easy as possible for anyone to spin up new, bespoke products and find demand if their work is good enough. The more live code that exists on Replit or that lives on a blockchain accessed via Replit, and the less dead code sitting in GitHub repositories, the more people will be able to remix the internet.
Replit is a special company in isolation, but it gets really interesting when you combine it with web3.
One of the big innovations of web3 is that there’s a shared backend and database on top of which anyone who knows how to code can build unique user experiences. In the ideal state, it could actually cause a reallocation of engineering resources globally to focus more on creating differentiated front-ends and less on re-building the same backend infrastructure and databases over and over again.
Combining that shift with a billion new software creators will have unimaginably dramatic effects on what type of software products people create. That’s the fun part about building legos: it’s impossible to predict exactly what will emerge, but easy to predict that we’re going to see a mind blowing amount of innovation. Each project is a jumping off point for the next.
The internet will only be as decentralized as people’s ability to code, or at least understand code, is democratized. Replit is building a world in which billions of people can build.
Let’s remix the internet. Set up your Replit and start hacking.
How did you like this week’s Not Boring? Your feedback helps me make this great.
Thanks for reading and see you on Monday,