Editorial front page
DevelopingAI-edited source brief

Cognition raises $1B at $25B valuation as AI coding agents go enterprise

The autonomous coding startup has more than doubled its valuation in eight months, drawing fresh capital even as Big Tech pushes its own programming tools.

Published Updated 6 sources0 Reddit5 web85% confidence

What matters

  • Cognition raised more than $1 billion at a $25 billion pre-money valuation, up from a $10.2 billion post-money valuation in September 2025.
  • The company claims a $492 million annualized revenue run rate and says Devin usage doubles every eight weeks.
  • Lux Capital and General Catalyst led the round, with participation from Founders Fund, 8VC, Ribbit Capital, Atreides, and Layer Global.
  • Enterprise customers include NASA, Goldman Sachs, Mercedes-Benz, the U.S. Army, and Palantir.
  • Cognition faces direct competition from OpenAI’s Codex, Anthropic’s Claude Code, and Google’s Jules.

Cognition, the startup behind the autonomous AI coding agent Devin, has closed a massive new funding round that vaults it into the top tier of private AI companies. The financing arrives as the company claims a nearly half-billion-dollar revenue run rate and a roster of blue-chip enterprise customers, even as the largest model makers build rival programming tools.

What happened

Cognition announced Wednesday that it has raised more than $1 billion at a $25 billion pre-money valuation. The round was led by Lux Capital and General Catalyst, with existing investors including Founders Fund and 8VC adding capital, and new backers Ribbit Capital, Atreides, and Layer Global joining.

The valuation is a sharp jump from the $10.2 billion post-money price tag Cognition hit just eight months ago, in September 2025, when it raised $400 million. The company said it has reached a $492 million annualized revenue run rate. Prior reports traced that figure from $1 million in September 2024 to $73 million by mid-2025, highlighting the speed of its recent growth.

Why it matters

The deal is a loud vote of confidence that standalone AI coding agents can carve out a durable market. OpenAI’s Codex, Anthropic’s Claude Code, and Google’s Jules are all competing for the same developer mindshare, and Google struck an acqui-hire deal with Windsurf last year before Cognition acquired the remaining assets. Despite that concentrated competition, Cognition has landed customers including Mercedes-Benz, NASA, Goldman Sachs, Santander, Citi, Dell, Cisco, Palantir, and the U.S. Army.

Cognition has tried to differentiate Devin as a fully autonomous agent that plans, writes, debugs, and deploys code, rather than a simple autocomplete assistant. The company says Devin’s performance on the SWE-Bench software-engineering benchmark has climbed from 13% to nearly 90%, and that usage doubles every eight weeks. As of March 2026, Cognition employed about 305 people.

Public reaction

No strong public signal was available from Reddit or broad social channels at press time. Earlier industry coverage noted the competitive-programming pedigree of CEO Scott Wu—a three-time International Olympiad in Informatics gold medalist—but widespread user sentiment remains thinly documented.

What to watch

The central question is whether Cognition can convert its current momentum into sustainable enterprise retention at a $25 billion price tag. With roughly 305 employees, the startup is valued at about $82 million per head, an extraordinary multiple that assumes years of hyper-growth.

Watch for how Devin performs against deeply integrated Big Tech alternatives, especially as OpenAI and Anthropic bundle coding agents with broader model platforms. Cognition’s ability to keep landing regulated, security-conscious customers like banks and government agencies may determine if independent AI coding tools remain a distinct category or get absorbed into larger suites.

Sources

Public reaction

No substantial Reddit or public forum discussion was captured for this funding announcement. Earlier coverage noted skepticism about whether autonomous coding agents could outperform Big Tech tools, but concrete user sentiment remains limited.

Signals

  • No strong public signal available

Open questions

  • Can Cognition sustain its revenue growth rate at enterprise scale?
  • How does Devin compare in real-world deployments against Claude Code and Codex?

What to do next

Developers

Evaluate Devin for autonomous task handling against Copilot or Codex for daily coding workflows.

Pricing and integration vary; test whether full autonomy reduces iteration time or introduces debugging overhead.

Founders

Study Cognition's enterprise sales motion into regulated industries.

Landing NASA, Goldman Sachs, and the U.S. Army suggests a playbook for selling high-visibility AI tools into compliance-heavy environments.

PMs

Benchmark internal coding assistants against Devin's SWE-Bench scores and security posture.

Enterprise buyers are increasingly using benchmark performance and compliance posture as procurement filters.

Investors

Treat the $25 billion valuation as a ceiling test for AI-agent revenue multiples.

At roughly 51× ARR, the round prices in extreme growth; compare retention and expansion data before assuming the multiple is repeatable.

Operators

Audit where autonomous coding agents can replace outsourced dev tasks or legacy automation.

A $492 million run rate implies real budget shifts inside large enterprises; map those use cases to your own cost centers.

Testing notes

Caveats

  • The story covers a private fundraising event and valuation update, not a new product release, API, or publicly available model. Developers can already access Devin through Cognition's platform, but no new testing surface was announced in this story.