Open a new tab. Type a sentence. A working app comes back. That is what 40 million people now do at replit.com - and the line between thinking about software and using it just got a lot shorter.
01 / WHO THEY ARE NOWThe browser is the IDE. The IDE is an agent.
Replit, in May 2026, is not the polite cloud editor it was when it slipped out of Y Combinator. It is a $9B company with 40M users, a CEO who tweets like a founder who has nothing left to prove, and a product that will - given a paragraph of plain English - return a deployed, working application with a URL you can text to your mother.
The Foster City headquarters is mostly symbolic. Most of the 450 employees live somewhere else. The product lives in your browser. The customers live everywhere - hobbyists in Lagos, founders in Bangalore, designers in Brooklyn, and increasingly, enterprise developers at Fortune 500s who were told by their CIOs that they should "look into Replit." The company doesn't sell developer tools so much as it sells the disappearance of them.
02 / THE PROBLEM THEY SAWSoftware was never short on programmers. It was short on editors.
For roughly forty years, building software required three things you do not have on a phone in a coffee shop - a laptop, a configured environment, and time. The industry treated this as physics rather than friction. Universities built degrees around it. Boot camps charged for it. Stack Overflow lived off it.
Replit's bet, made unfashionably early, was that the configuration step was the entire problem. Strip away the install, the dotfiles, the missing dependency, the wrong Python version - and what you have left is not a developer shortage. You have a permission shortage. The world has plenty of people with ideas. It just doesn't let most of them ship.
That belief sounded charming in 2016, when "code in your browser" was the kind of thing you said at a hackathon. It sounded delusional in 2020, when serious engineers still treated VS Code on a MacBook as the only adult way to work. It started sounding inevitable in 2024, when Replit Agent began writing - and shipping - the apps for you.
03 / THE FOUNDERS' BETA Jordanian engineer, a designer, and a brother walk into a browser.
Amjad Masad grew up in Amman, the son of refugees, with a civil-engineer father and a stay-at-home mother. He learned to program from a pirated CD. He later joined Codecademy as a founding engineer, then Facebook, where he worked on React Native. None of which prepared him, exactly, for trying to convince the venture-capital industry that the future of coding was a tab.
He started Replit in 2016 with his brother Faris Masad and the designer Haya Odeh - now his wife and the company's head of design. The original product was a "repl" - a read-eval-print loop in the browser. It was small, useful, and unmistakably a developer tool for developers. It also happened to be the perfect substrate for everything that came next.
The bet looked patient for a long time. Y Combinator backed them in 2018. a16z led the seed. The company spent five years building infrastructure most people couldn't see: live multiplayer cursors, container orchestration, package management, deploy targets, version control, an AI-friendly file system. Boring decisions, made slowly, that would later turn out to be load-bearing.
The Replit Calendar
04 / THE PRODUCTOne prompt. One URL. One deployed thing.
Strip out the marketing and Replit does four things. It hosts your code. It runs your code. It collaborates on your code. And, now, it writes your code. The fourth piece - Replit Agent - is what changed the economics. Earlier coding assistants would autocomplete a line. Agent will autocomplete a company.
05 / THE PROOFCharts that no longer require a caveat.
For most of Replit's life the right thing to do, as a journalist, was to nod politely at the user count and ask how it converts to revenue. That question has stopped being interesting. According to Sacra, the company went from $2.8M in annualized revenue at the start of 2025 to $150M by September. Full-year 2025 revenue landed at $240M. The 2026 stated target is $1B in run-rate.
Annualized Revenue, in millions
The customer list grew up too. Replit is increasingly the AI prototyping environment of choice for serious shops - Google, Anthropic, Databricks, Accenture, plus a long tail of startups whose founders are technically non-technical and unbothered about saying so. Anthropic's Claude models power much of the Agent's reasoning. Google Cloud carries the infrastructure. Databricks joined the Series D - rare for a company that has its own developer surface.
06 / THE MISSIONThe next billion developers won't call themselves developers.
The phrasing on the website is "empower the next billion software creators." The phrasing in private, around the office, is less polished and more interesting - Replit thinks the word "developer" is about to mean roughly what the word "writer" means today. Most people who write do not call themselves writers. Most people who will, soon, make software, will not call themselves developers.
That is a strategic position, not just a slogan. It explains why Replit shipped a mobile app for coding before most professional coders thought a mobile coding app was a serious idea. It explains why Agent does not assume you know what a dependency is. It explains why the company is fine with the term "vibe coding" - a phrase that makes traditional engineers visibly uncomfortable and accurately describes what most new users are doing.
07 / WHY IT MATTERS TOMORROWSoftware stops being an industry. It becomes an action.
If Replit is wrong, the company is still a useful cloud IDE with a respectable enterprise practice and a strong AI feature. The downside, by venture standards, is fine. If Replit is right, the consequences are more interesting - and more uncomfortable. The cost of producing a single application falls toward zero. The supply of applications goes up by orders of magnitude. The bottleneck moves from engineers to taste, distribution, and trust. The shape of every adjacent industry - cloud, SaaS, IT services, even computer-science education - rearranges itself around that fact.
That is the version of the future Amjad Masad has been describing on podcasts for two years. Until 2024, the future sounded speculative. After Agent shipped, it sounded predictive. After the Series D priced at $9B, it started sounding like the working assumption.
None of which guarantees Replit wins. Agentic coding is now a contested category - GitHub, Anthropic, OpenAI, and a fistful of startups all want this real estate. But Replit has something the new entrants don't: a decade of infrastructure built before AI made it look obvious, and a user base that already trusts the browser tab.
Watch & listen
The links
08 / RETURN TO THE OPENINGOpen a new tab. Try it now.
That sentence at the very top - open a new tab, type something, get an app - was a parlor trick in 2016. It was a product demo in 2024. In 2026, it is the default behavior of forty million people, and the founders who said it would happen are now responsible for the company doing it.
Replit did not win because it built the best editor. It won, so far, because it spent eight years acting on a premise everyone else was happy to dismiss. Software shouldn't require a setup. The browser is enough. Most people who could be making things are not making things, and that is fixable. None of those statements are clever. They are just, eventually, true.