Replit CEO Amjad Masad says the company is building what he calls a self-driving company, and the numbers are moving fast enough to test the thesis. Replit raised $400 million in March 2026 at a $9 billion valuation — triple its worth six months earlier — and says it is on track for $1 billion in annual recurring revenue by year-end. That is up from $2.8 million in revenue in all of 2024.
More than 50 million people now build on Replit, including 85 percent of the Fortune 500. Last month the company published a claim that its engineers nearly tripled the amount of code each person ships in six months, while holding code quality, after rolling agents through the company. Replit has begun canceling its own software contracts along the way, replacing most of its analytics vendors with tools its employees built in-house.
Masad, 38, co-founded Replit in 2016 with his wife, the designer Haya Odeh, after stints as a founding engineer at Codecademy and running JavaScript infrastructure at Facebook. He grew up in Amman, Jordan, and frames the company's mission as widening access to software creation. Coding, he argues, has historically been available to roughly 0.5 percent of the human population — those who could afford a computer science degree.
“This technology has always been accessible to 0.5% of the human population, those who could afford a computer science degree.”— Amjad Masad, Co-founder and CEO of Replit
Key facts
- 01Replit raised $400M in March 2026 at a $9B valuation, triple its mark from six months earlier.
- 02The company is targeting $1B in ARR by end of 2026, up from $2.8M in total 2024 revenue.
- 03More than 50M people build on Replit, including 85% of the Fortune 500.
- 04Internal engineers nearly tripled per-person code shipped in six months after agent rollout.
- 05CEO Amjad Masad says Moonshot AI's Kimi K3 is now a better designer than [Claude](/claude).
That framing is why Masad hates the term vibe coding, even though Replit is one of the biggest names associated with it. Type a sentence into Replit today and an agent designs, writes, tests, deploys, and hosts the application. Masad said he now tells people not to bother learning to code at all — a striking reversal after 15 years spent making coding easier to learn, first at Codecademy and then at Replit.
The internal operating model is where the self-driving pitch gets concrete. Masad describes hiring engineers who "only look at the code when it's about to get deployed," running an internal bot as "the brain of the company," and calling a CEO "a kind of glorified router" whose routing should mostly be automated. It is a management philosophy explicitly designed to shrink headcount per unit of output.
Masad is unusually direct about what that means for jobs. He said companies will get smaller and will do layoffs — "100 percent sure" — even as more companies get created in aggregate. He called it cruel to tell workers in their 50s to retrain, and said that is a job for a government he does not think is engaging with the problem.
“Net net, there's going to be more jobs, but also there's going to be more companies.”— Amjad Masad, Co-founder and CEO of Replit
The product surface is widening too. Replit last month launched Replit Design, an agent-based creative tool, and Masad said Moonshot AI's Kimi K3 model, available inside Replit Design, is now a better designer than Anthropic's Claude. That is a pointed claim from a CEO whose company is one of Anthropic's larger enterprise customers, and it signals that Replit is willing to route work to whichever model wins a given task.
Asked to grade a prediction from Raycast's Thomas Paul Mann that within three years half the software on a user's computer will be apps they built themselves, Masad went further — and toward a conclusion that arguably undermines Replit's own business. He argued that in three years users will run fewer apps and more agents, with those agents calling the remaining apps on the user's behalf. The app, in that telling, becomes an intermediate artifact that agents increasingly skip.
“In three years, we'll be using fewer apps, we'll be using more agents, and those agents will be using the apps on our behalf.”— Amjad Masad, Co-founder and CEO of Replit
Masad's own experiments illustrate the pitch. He said he trained a fully neural-network chess engine — an unsolved problem in the literature, since most engines like AlphaGo and AlphaZero are hybrids — and posted it on Lichess with a 1300 Elo score, or roughly 1700 against humans. His point is not the chess result but the leverage: a non-researcher shipping a research-grade artifact by directing an agent, the way, in his framing, Neo learns to fly a helicopter in The Matrix.
The counterweight is that Replit's growth math still has to survive a market where every coding-agent competitor is compressing prices, where enterprise buyers have started asking hard questions about code quality behind the productivity numbers, and where Masad's own three-year forecast implies the app-building layer he sells gets bypassed. Tripling per-engineer output is a striking internal metric but it is self-reported, and the $1 billion ARR target implies revenue growth roughly 350 times faster than 2024 — a pace that leaves little margin for a slowdown in agentic-coding demand.
Replit's real bet is that the ceiling on non-developer software creation is far higher than the market has priced in, and that whoever owns the default surface for agent-built software captures a category that did not exist two years ago. If Masad is right that agents will consume apps rather than users consuming them, the winners in this cycle will be the platforms that host the agents, sell the compute, and price the transactions between them — and Replit is trying to become one of those platforms before the app layer thins out underneath it.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
