Cognition, the startup that developed the Devin coding assistant, announced that it raised $2 billion in a funding round that valued the company at $48 billion. Andreessen Horowitz, Accel, Founders Fund, General Catalyst, and Avenir led the round, just four months after a previous round valued the company at $26 billion.
The jump comes as a growing number of companies compete for the market for AI-powered coding assistants, one of the technology’s most prominent applications for businesses and developers. Cognition said its annualized revenue run rate rose from $492 million when it announced its previous round in May to $900 million currently.
However, the company did not explain how it calculated this figure. The term annualized revenue run rate is typically used to refer to multiplying the revenue from one month by 12, making it an indicator of the current pace rather than necessarily the revenue actually generated over a full year.
What Does the Valuation Reveal About the Market?
The round indicates that investors do not view the AI coding market as one in which a single company will capture most of the value. Cognition’s high valuation, despite its reported revenue being lower than that of its competitor Cursor, reflects continued confidence that several platforms can achieve significant market shares.
Cursor had discussed raising funding at a $50 billion valuation in April, before agreeing that same month to be sold to SpaceX for $60 billion. At the time, its annualized revenue run rate had exceeded $2 billion. Based on the disclosed figures, Cognition is receiving a higher valuation relative to its revenue than the valuation Cursor was seeking before the deal.
Computing Costs Remain a Key Test
A high valuation does not guarantee that growth will translate into profitability. According to investors cited in earlier reports, Cursor turned to the SpaceX deal partly because of severe constraints on computing capacity. It remains unclear whether Cognition will face the same problem.
Cognition leases a server cluster from Nvidia at a cost of up to hundreds of millions of dollars annually, which could raise its total cash consumption to $800 million this year, according to The Information. The company is also working to train its own model using open-source alternatives, a move that could gradually reduce its reliance on the costly models of OpenAI and Anthropic and bring it closer to break-even, according to the source.
What Is Changing in Practice?
The deal shows that competition among coding assistants is not only about product quality, but also about the ability to finance infrastructure and run models at scale. The Information expects Cognition to reach an annualized revenue run rate of between $4 billion and $5 billion by the end of 2026, while Cursor was on track to exceed $6 billion by the end of the year, according to an earlier TechCrunch report.
The round carries additional significance because Andreessen Horowitz, one of Cursor’s principal backers and a beneficiary of its sale to SpaceX, has returned to lead financing for its competitor. Cognition’s enterprise customer base includes Mercedes-Benz, NASA, Goldman Sachs, and Citi. Nevertheless, the method used to calculate revenue, the actual scale of computing expenditures, and the company’s ability to reduce its reliance on external models remain open questions that will determine whether the valuation reflects sustainable growth or an expensive race to scale.