Replit hit $125M revenue in 2025. Now its suppliers are its rivals.
Replit generated over $125 million in revenue in 2025. Its founders say it’s on track for $1 billion in ARR by end of 2026. That’s the headline. The harder story is what comes next: the AI labs that power Replit’s core product are now building direct competitors to it.
This is the situation Amjad Masad and Haya Odeh find themselves in after nearly 15 years of building together, first as co-workers arguing about whether engineering or design should lead a product, then as partners, co-founders, and spouses.
Where It Started
Masad and Odeh met at Boss Consulting SA, a web consultancy in Amman, Jordan. Masad believed engineering should be the source of truth for software. Odeh argued UX should lead. The friction was productive. When Masad left the firm in 2009, they exchanged numbers at her birthday party and started going out.
Early in the relationship, Masad asked Odeh to help with a side project he’d started in college: a browser-based coding environment called Repl.it. The concept was simple: remove the friction of setting up a programming environment and deploying code, similar to how Google Docs brought word processing to the browser. Odeh designed the logo and worked on the interface. She didn’t fully understand the technology. She helped anyway.
The Long Road to Product-Market Fit
The couple moved to the US in the early 2010s. Masad took a job at Codecademy in New York, then moved to Facebook in Silicon Valley. Odeh built a career in internationalization design and started growing Replit’s user base on the side, taking it from around 5,000 to 50,000 accounts.
In 2016, they went all in. Masad quit Facebook. They officially founded the company. Their original thesis was a classroom product for teaching coding. It gained traction but wasn’t scalable. What they noticed instead was that users were building games and small apps in the browser coding environment without being asked to.
Replit applied to Y Combinator four times. On the fourth try, in winter 2018, it got in. Sam Altman, then YC’s president, had reached out to Masad after Paul Graham discovered Replit on Hacker News. In October 2018, Replit announced a $4.5 million raise led by Andreessen Horowitz. The new mission: a platform where anyone with minimal experience could build, ship, and acquire users for apps, all in one place.
The AI Pivot That Changed Everything
In late 2020, Masad got early access to GPT-3 and immediately understood what it meant for programming. At the time the model could write only a single line of code. He saw the trajectory anyway. In 2021, Replit raised $100 million from A.Capital and Coatue to hire AI engineers and build new features. The first major output was Ghostwriter, an AI autocomplete tool for partially written code.
Ghostwriter doubled Replit’s revenue. It still wasn’t enough. Masad had promised the board $100 million in ARR by end of 2023. The company had overhired. Masad and Odeh laid off 30 of roughly 120 employees. In an internal email Masad posted publicly, he wrote that “now is the time to move fast because the opportunity for Replit has never been clearer.”
The leaner team shipped Replit Agent in September 2024. Unlike Ghostwriter, Agent could build and iterate on apps through a chat interface, then handle deployment and hosting as well. It was built on Anthropic’s Claude 3.5 Sonnet model, which Masad credits with exhibiting significantly better coding capabilities than any model available before it. AI researcher Andrej Karpathy recommended the tool publicly on X. Masad says Replit was the first programming agent to market, and that it indirectly inspired Codex, Claude Code, Bolt, and Lovable.
“The researchers from OpenAI and Anthropic reached out to ask us, ‘How are you able to do this? We didn’t know the models were capable of that.'” — Amjad Masad
Replit’s ARR jumped from $4.1 million to over $250 million in roughly a year. Paid plans start at $18 a month. Karpathy gave the broader category a name in 2025: vibe coding.
What a Replit User Built With It
Eman Khoubian, a former real estate investor, saw Wispr Flow raise a $30 million Series A in June 2025 and decided he could build something similar. He used Replit to develop a web app that converts audio to text, named it Whisper AI, and bought the domain whisperai.com in October 2025. After the domain purchase, signups accelerated. He quit his day job. Whisper AI generated nearly $150,000 in 2025 and, as of mid-2026, was generating six figures in recurring revenue monthly.
Khoubian’s honest assessment of his biggest churn driver: users sign up thinking they’ve found Wispr Flow. Brand confusion is his primary retention problem. He’s building Chrome extensions and desktop apps to compete more directly. The moat problem, at a smaller scale, mirrors Replit’s own.
The Competitive Threat Replit Can’t Ignore
Replit competes with Lovable and Bolt on one front. On another front, it’s competing with the same AI labs it depends on for model access. Claude Code has grown to over $2.5 billion in ARR in a single year. OpenAI’s Codex reports more than five million weekly active users.
The structural disadvantage is real. Research firm SemiAnalysis reports that a $20 Claude Pro or ChatGPT Plus subscription covers $400 to $700 worth of tokens respectively. Replit pays full API token prices. Claude Code and Codex users get far more tokens per dollar than Replit subscribers do. Masad calls this out directly, framing it alongside Apple’s decision to block Replit’s iOS app updates for several months over a dispute about running code that could affect how apps function on the platform. He wrote on X in May that Replit had “worked things out with Apple” and released an update.
“Platforms should not preference themselves,” Masad says. “We should compete on the best product and the best service.”
The Moat Argument
Masad’s case for Replit’s defensibility rests on two things. First, design. Odeh leads a user experience built around how non-technical people actually think: chat for those who talk through ideas, a digital canvas for those who sketch visually. She says the long-term goal is for Replit’s UI to feel more like a mood board than a chatbot. Second, data. Georgian’s Margaret Wu, whose firm led Replit’s $400 million Series D in March, points to Replit’s store of metadata on user intent, including the prompts and questions users send to their agents. That data could train future models to be better at vibe coding specifically.
David Yoffie, a professor at Harvard Business School who studies platform strategy, frames the challenge starkly. Foundation model providers now occupy a position similar to Microsoft’s in the early 2000s: a powerful platform identifying high-revenue opportunities and incorporating them into the core product. Companies that survived Microsoft’s dominance, like Intuit and Adobe, did so by offering products that were hard to replicate. Yoffie says entrepreneurs with a short-term advantage might also want to consider selling to a model provider while the market is still hot.
Masad isn’t selling. He received a medal from Jordan’s King Abdullah II at the country’s 80th independence day celebrations in May, recognition that landed with weight for someone who grew up with, as he put it, “very little means.” The ambition that got him here isn’t going anywhere. “You can’t rest,” he says. “If you miss a beat, someone else will take it.”
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
