TL;DR
- Funding Round: Cognition AI has raised more than $1 billion at a $26 billion valuation behind its Devin coding agent.
- Traction Signal: Cognition’s reported customer and usage figures appear to have pushed Devin’s annualized revenue to about $492 million after six months of rapid growth.
- Competitive Test: Cognition now has to widen paid deployment while Anthropic and OpenAI keep expanding rival coding-agent access and workflow tools.
Cognition AI has raised more than $1 billion at a $26 billion valuation, giving the AI coding startup behind the Devin AI agent system one of the biggest new cash injections in the coding-agent market.
Investors are treating AI coding tools as businesses that may hold enterprise budgets over time, not just as a short-lived model boom. Round leaders included Lux Capital, General Catalyst and 8VC, with Ribbit Capital, Atreides and Layer Global also joining.
Cognition builds Devin, an autonomous AI software engineer. Reported operating traction, not just the financing headline, is doing much of the work behind the new valuation.
Why Investors Are Paying Up for Devin
Reported operating figures extend the financing case into day-to-day use. Cognition’s reported customer and usage figures have pushed annualized revenue to about $492 million after six months of 50 percent month-over-month corporate growth. Because a run-rate measures revenue at the current pace over a full year, it gives investors a rough benchmark for how quickly paid deployment may be scaling.
Customer breadth adds another layer to that revenue story. Mercedes-Benz, NASA, Goldman Sachs and Santander appear in Cognition’s reported customer roster, while enterprise usage increased tenfold since the start of 2026. Together, those signals tie the valuation to repeat enterprise spending rather than novelty alone.
Official company material reinforces the deployment picture. According to the company, Devin was already working inside thousands of companies by 2025, suggesting the product has moved well beyond small pilots before this round closed. Scale on that level implies Cognition had enough live usage to surface bugs, workflow limits, and procurement friction before asking investors to pay up.
Cognition also named Goldman Sachs, Santander, and Nubank among the engineering teams using Devin. Together, those names point to interest from both established financial firms and fast-moving digital banks that already run sizable software operations.
Pull-request volume adds a separate operational scale signal. Cognition said last November Devin by then had already merged hundreds of thousands of pull requests.
One reported benchmark put Devin at eight days instead of eight months for a modernization project, which helps explain why backers were willing to support such a sharp valuation jump.
How Fast Cognition’s Position Has Changed
The new valuation followed a much lower benchmark in September 2025. Cognition then was valued at $10.2 billion after a September 2025 funding round, so the new valuation arrived in less than a year.
Devin’s rollout had already widened before the latest round closed. Cognition had made Devin generally available for teams in December 2024.
Self-serve pricing and the earlier Team entry point broadened the path from a small test to a wider engineering rollout, while the earlier Windsurf acquisition gave Cognition another installed base to target.
What the Money Has to Prove Next
Competitive pressure now creates the execution test. Cognition plans to use the new capital to widen product adoption and hire more employees. Success now depends on turning a sizable raise into broader paid deployment inside large engineering teams, where procurement reviews, security checks, and workflow fit often slow expansion even after a promising pilot.
Competition is not slowing down. Anthropic recently doubled Claude Code’s five-hour rate limits and removed peak-hours limits for some paid plans. Lower limits reduce one practical barrier for teams that want heavier coding-agent use during busy work periods. Easier access also gives buyers more room to compare coding tools in real workflows before they commit to a broader rollout.
OpenAI’s workflow push is advancing in parallel. It published an open-source spec for Codex orchestration: Symphony for isolated workspaces tied to issue-tracker tasks. By formalizing that structure, the design pushes the category toward managed, repeatable workflows instead of one-off prompt sessions, raising the bar for how enterprise buyers may judge tools like Devin.
