While AI coding startups like Cursor close brow-raising rounds on barely three years of existence, Replit’s path to a $3 billion valuation has been anything but swift. For CEO Amjad Masad, who’s been building tools to democratize programming since 2009, it’s a story of muscling through multiple failed business models, years stuck at the same revenue plateau, and a reckoning last year that forced him to cut half his staff.

That makes what happened next more remarkable. Earlier this month, the Bay Area-based company closed a $250 million funding round led by Prysm Capital, nearly tripling its valuation from 2023. The raise came on the heels of never-before-seen revenue growth for the company — from just $2.8 million last year to $150 million in annualized revenue in less than a year. But for Masad, this moment represents something more than finally realizing financial traction. It’s the culmination of a 16-year obsession.

“Our mission has always been the same,” Masad told me on the newest episode of TechCrunch’s StrictlyVC Download podcast. “Initially, we said we want to make programming more accessible, and then we sort of upped the ante a little bit. We said we’re going to create a billion programmers.”

It’s purposely audacious – what a headline! – but it’s also something that Masad, a Palestinian-Jordanian, has been working toward for his entire career. As he tells it, he came to the United States in 2012 after his open-source coding project began gaining attention – including catching the eye of the New York Times. But he’d been making programming more accessible since building his first online coding experience back in 2009, with his work as an early engineer at the startup Codecademy kicking off what became the massively online open courses (MOOC) revolution. (His code also powered the in-browser tutorials of Udacity, a Codecademy rival that launched in 2012, one year after Codecademy was founded.)

Still, turning that vision into a viable business of his own proved a lot harder than he anticipated. Replit was founded in 2016, and for eight long years, the company struggled to find product-market fit. “We had reached that $2.83 million [in annual recurring revenue] back in ’21, maybe,” Masad recalled. “And so this is how painful it’s been. We’ve been hovering around the same revenue for like four or five years.”

The company tried selling to schools (“incredibly difficult,” Masad noted), cycling through different business models, and watched each one stabilize around the same modest revenue level.

Along the way, Replit built sophisticated infrastructure for cloud development environments and “multiplayer coding,” collaborative editing akin to Google Docs but for programming. But the technical achievement wasn’t translating into revenue growth, and by last year, with the company at 130 employees and burning through cash, Masad said he had to make a painful decision. “I looked at our burn, and I looked at our progress on our revenue chart, and it just didn’t make any sense. The business wasn’t viable.” Replit cut its headcount by 50%, bringing it down to around 60 to 70 people at its lowest point.

Then came the breakthrough.

Last fall, Replit launched Replit Agent, which Masad calls “the first agent-based coding experience in the world” that can’t just write code but “debug it, deploy, provision the database for you, just act as a true software engineering partner.”

Soon after, in January of this year, he announced that Replit was abandoning professional developers as its core market.

“Hacker News was really unhappy,” Masad acknowledged when we talked. But he also hasn’t looked back, completely moving away from competing in the crowded market of tools for professional developers – where companies like Cursor, GitHub Copilot, and others are battling it out – to instead focus on creating a billion software developers from white-collar employees with no technical background.

“The idea of making programming more accessible to the average individual, to the knowledge worker, really, that’s where we think our market is,” Masad explains. “It’s a fundamentally new market.”

Right now, that bet looks very smart. Numerous reports this summer said that revenue at Replit had grown to over $150 million in annualized revenue and Masad hinted that it’s now even higher. He also said that unlike many AI-powered coding companies, Replit is gross margin positive. On enterprise deals, which make up an increasing share of revenue, margins are “80% to 90%,” according to Masad.

It’s hard to verify such a claim, but Replit’s market position received some validation this week when Andreessen Horowitz released its first AI Spending Report in partnership with fintech firm Mercury. Analyzing transaction data from Mercury, the report tracked the top 50 AI-native application layer companies that startups are actually spending money on. While major labs OpenAI and Anthropic took the top two spots, Replit landed at No. 3, outranking every other development tool. (Worth noting: Andreessen Horowitz has invested in multiple rounds of funding for Replit.)

Profitability is rare in AI coding because many competitors face what Masad calls “the negative gross margin trap.” The reality is that serving professional developers with AI assistance can be compute-intensive. Counterintuitively, Replit’s focus on non-technical users – who might seem like they’d require more AI assistance – works in their favor on the business model front for enterprise customers like Zillow, Duolingo, and Coinbase, which pay $100 per seat, plus usage-based pricing built on top.

This new path hasn’t been without some faceplants. In July, venture capitalist Jason Lemkin went viral after the newest version of Replit’s AI agent deleted his production database with 100-plus executive contacts, fabricating 4,000 fake records afterward and later admitting to Lemkin that it “panicked.” (There is a failure mode in AI agents called reward hacking, where models become so obsessed with achieving a certain goal that they effectively cheat when they miss the mark.)

Rather than becoming defensive, Masad and his team owned the problem. In fact, says Masad, within two days, they rolled out an automatic safety system that separates a user’s “practice” database from their “real” one. The way Masad describes it, it’s a little like having two versions of a website’s filing cabinet — the AI agent can experiment freely in a development database, but the production database, which is the real thing that users interact with, is completely walled off.

Masad told me the incident ultimately put the company on stronger footing, given the problems around safety and security it needed to figure out — and fast. “If you solve hard problems, then you have a technology moat,” he said. (Lemkin, for his part, says he has become a super user of Replit despite having no technical background just months ago.)

Still, even now, Replit isn’t out of the woods. If anything, its success has painted a target on its back. To wit, the company — which now employs 110 people — still faces an existential threat from the very AI labs whose models power its platform: Anthropic and OpenAI. Both companies have launched their own coding tools that compete directly with companies like Replit and Cursor, and these foundation model companies can afford to subsidize their coding tools and post-train their models on their own products, optimizing performance in ways that third-party platforms might always struggle to replicate.

Replit’s advantage, according to Masad, lies in targeting non-technical users rather than professional developers, plus the sophisticated infrastructure around deployment and database management that it has built and which foundation model companies still don’t prioritize (for now).

Plus, Replit has another unusual advantage for a startup: a $350 million war chest. Despite raising $100 million in 2023, the company “hadn’t touched” those funds by the time it raised this latest round, Masad told me. The company is capital efficient by design, though Masad joked that as an entrepreneur who grew up watching his refugee father struggle, “one thing I need to learn is to be less frugal and start spending money.”

Whether that edge keeps Replit ahead of competitors is an open question, and it’s one about which Masad is mindful. Right now, the plan is to scale operations, accelerate product development, and pursue acquisitions — both acqui-hires and potentially companies working on agent automation in specific verticals. But for Masad, who appeared on Joe Rogan’s podcast in July and has seen his company’s fortunes transform, the moment is bittersweet. When asked how it feels to be receiving so much attention – not to mention that $3 billion valuation – he invoked the adage that “this too shall pass. This might mean that when you’re in a bad situation, that’ll pass, but we’re also in a good situation that will pass.”

It’s a stoic response from someone who spent the better part of a decade working away at the same revenue level, convinced that AI agents would eventually transform programming but unable to prove it to the market. But one major difference between Replit and the wave of AI coding startups now flooding the market is that Masad has lived through multiple hype cycles and has he emerged with something relatively differentiated – and reportedly profitable.

“I’ve learned to be a little stoic,” he said. “What matters is for us to do the right thing, be principled, and move forward.”