Analysis
Cognition, the AI coding startup behind the autonomous software engineer Devin, is in early talks with investors for a new funding round that could value the company at $40 billion or more, according to Bloomberg. That would mark more than a 50% jump from the $26 billion valuation Cognition set in May, when it raised $1 billion led by Lux Capital, General Catalyst and 8VC, with Peter Thiel's Founders Fund, Elad Gil and Ribbit Capital also participating. TechCrunch reports the terms -- lead investor, round size, final price -- remain unconfirmed.
Founded in 2023 by Scott Wu, Cognition built its early reputation on Devin, marketed as an autonomous AI software engineer capable of independently writing, testing and shipping code rather than just suggesting it. The company's valuation has compounded fast even by 2026 AI-startup standards: $10.2 billion two months after acquiring Windsurf last September, $26 billion after May's $1 billion raise, and now a prospective $40 billion-plus mark less than three months later.
The competitive field Cognition is racing against has only gotten more crowded: GitHub Copilot (Microsoft), Cursor -- now owned by SpaceX in a separate $60 billion deal -- and Anthropic's own Claude Code are all fighting for the same enterprise engineering budget, and each has a distribution advantage Cognition lacks, whether that's Microsoft's developer install base, SpaceX's GPU fleet, or Anthropic's frontier-model relationship. Cognition's Windsurf acquisition, made after Windsurf's earlier talks with Google fell apart, gave it a second product surface and additional enterprise customers, which is reportedly part of the case for the higher mark.
The number underwriting the round is revenue: Cognition is said to be approaching a $1 billion annualized run rate, up sharply from the roughly $492 million ARR the company was reported at earlier this year. If accurate, that would put a $40 billion valuation at roughly 40x forward revenue -- aggressive, though not unusual for a category where investors are pricing multi-year growth curves rather than trailing revenue.
The risk sits in how much of that revenue is durable, coding-specific spend versus one-time enterprise pilots that don't renew once a company's engineering team settles on a single AI coding vendor. Coding-tool spend has proven stickier than most AI categories so far, but the field has three well-capitalized entrants now backed by hyperscaler-scale balance sheets, and Cognition's next raise -- if it closes -- will be a bet that Devin's autonomy claim keeps winning enterprise deals against competitors with deeper pockets.
Cognition's Windsurf deal is itself a reminder of how fast this category has consolidated: Windsurf was originally in advanced acquisition talks with OpenAI before that deal collapsed, after which Google licensed Windsurf's core technology and hired away several of its top researchers, leaving Cognition to acquire what remained of the company and its enterprise customer base last July. That history is part of why investors are willing to underwrite Cognition's valuation jumps in months rather than years -- the coding-AI land grab has already claimed one near-miss, and nobody wants to be the investor who passed on the round before a company locks in category leadership.
What to watch next is whether Cognition actually closes a round at $40 billion or settles lower the way Groq did after its own Nvidia-driven reset -- talk-stage valuations in this cycle have not always survived contact with a term sheet.
