Cognition announced a $2 billion fundraising at a $48 billion valuation for its coding assistant, Devin. The round was led by Andreessen Horowitz, Accel, Founders Fund, General Catalyst and Avenir, and follows a prior raise four months earlier at a $26 billion valuation.
Revenue and market position
The company reported that its annualized run-rate revenue grew from $492 million to $900 million since its May raise. Cognition did not disclose how it calculates that run rate; the metric is commonly defined as one month’s revenue multiplied by 12.
Investors see the result as evidence that AI coding remains a market with room for multiple large players rather than a winner-take-all outcome. In related activity earlier in the year, Cursor was in talks in April to raise at a $50 billion valuation and later agreed to sell to SpaceX for $60 billion; at the time Cursor’s annualized revenue had surpassed $2 billion. That sequence implies Cognition currently carries a higher revenue multiple than Cursor did in the spring.
Costs, compute and product strategy
Challenges around compute capacity remain relevant. Cognition leases an Nvidia server cluster described as costing hundreds of millions of dollars per year and could face total cash burn of about $800 million this year, according to reporting by The Information. Investors familiar with Cursor’s financials said compute constraints were a major factor in Cursor’s sale to SpaceX.
Cognition is training its own model using open source alternatives, a strategy the company says will reduce reliance on third-party models from OpenAI and Anthropic and help lower costs over time. The Information projects Cognition could reach $4 billion to $5 billion in annualized revenue by the end of 2026; by comparison, earlier reporting indicated Cursor was on pace to surpass $6 billion by year-end.
Founded in 2024 by Scott Wu, Cognition counts Mercedes-Benz, NASA, Goldman Sachs and Citi among its enterprise customers. The new financing reflects investor appetite for multiple contenders in AI coding tools, even as compute expenses and model development shape each company’s trajectory.
Original source: TechCrunch AI