- Cognition raised over $1 billion at a valuation north of $26 billion — more than double the $10 billion mark from September 2025.
- Lux Capital, General Catalyst, and 8VC led the round; Founders Fund and Ribbit Capital participated.
- Cognition reports $492 million annualized revenue and 10x enterprise usage growth this year. Clients include Citi, Mercedes-Benz, Goldman Sachs, Dell, and the U.S. military.
- 89% of Cognition’s own code is now written by Devin; the company also released its own model SWE-1.6.
What Happened
Cognition — the company behind AI software developer Devin — raised over $1 billion at a valuation north of $26 billion, The Decoder reported. The round more than doubles Cognition’s $10 billion valuation from September 2025, when Devin had raised $400 million. Lux Capital, General Catalyst, and 8VC led the round; Founders Fund and Ribbit Capital joined.
Why It Matters
The $26 billion mark places Cognition in the upper tier of pre-IPO AI companies — alongside xAI, Anthropic, Mistral, and Together AI on valuation, though substantially below OpenAI’s $852 billion. The doubling in under nine months reflects investor confidence in AI coding agents specifically, even as a vocal contrarian camp questions whether the category has product-market fit at the prices being paid.
Cognition’s commercial traction is genuinely substantial: $492 million annualized revenue with 10x enterprise usage growth year-over-year, and an enterprise client base spanning Citi, Mercedes-Benz, Goldman Sachs, Dell, and the U.S. military. The customer mix is heavily weighted toward regulated industries (financial services, defense, automotive) where AI-augmented engineering would historically have been a tough sell.
Technical Details
Cognition’s Devin remains the company’s flagship AI software developer. The company works model-agnostically with different AI labs and recently released its own model, SWE-1.6. Internally, 89 percent of Cognition’s code is now written by Devin — a metric that mirrors Anthropic’s claim that ‘most software at Anthropic is now written by Claude’ from the Code with Claude conference last week.
The competitive landscape includes open-source alternatives. SWE-Agent is an open-source alternative to Devin. Cursor’s Composer 2.5 release on May 17 — matching Claude Opus 4.7 and GPT-5.5 on SWE-Bench Multilingual at one-tenth the cost — represents the cheap-inference variant. The contrarian view, voiced by developer George Hotz: AI coding agents are ‘the most costly mistakes’ in software development history.
Who’s Affected
Cognition’s investors gain a step-up valuation marker. Cognition’s enterprise clients (Citi, Mercedes-Benz, Goldman Sachs, Dell, US military) gain confidence in continued product investment. Anthropic Claude Code and OpenAI Codex face a directly-competitive coding-agent platform at premium pricing. Open-source alternatives (SWE-Agent, the Aider community) gain visibility from the funding event. AI venture-capital firms continue to bet heavily on the coding-agent category; the next 12-24 months will determine whether the $26 billion valuation is justified by revenue ramp.
What’s Next
Cognition has not announced an IPO timeline. The company has signalled it intends to remain independent and continue model investment via SWE-series releases. The broader debate about coding-agent product-market fit — between the $26B valuation case and the ‘most costly mistakes’ contrarian case — will resolve incrementally through 2026-2027 as enterprise renewal rates and revenue ramp become visible.
