Assign a coding task such as a refactor or migration and have the agent carry it through.
Cognition Funding History
Cognition develops AI tools for software engineering, including the Devin coding agent and the Windsurf development environment. Devin works across an IDE, shell and browser to plan, write, debug and deploy code, while the company serves enterprise engineering teams.
Company Overview
Founded 2023 · San Francisco, CA, USA
Official WebsiteCognition Funding Timeline
Series E
Over $2 billion · $48B valuation
Cognition reported raising more than $2 billion at a $48 billion valuation.
Led by Andreessen Horowitz, Accel. Participation included Benchmark, Bessemer, Kleiner Perkins, Greylock, Lightspeed Venture Partners, Altimeter Capital, BOND, Meritech Capital, Atreides, T. Rowe Price, Lux Capital, 8VC, D1 Capital, DST Global, Positive Sum, BoxGroup, Ribbit Capital, Swish VC, Definition Capital, Hanabi Capital, Bain Capital Ventures, Layer Global, Alpha Wave, Diffusion, Founders Fund, General Catalyst, Avenir.
Source · cognition.com ↗Series D
$1B · $26B valuation
Alongside the financing, Cognition reported that its revenue run rate had reached $492 million and that enterprise usage had grown more than tenfold since the start of the year.
Led by Lux Capital, General Catalyst, 8VC. Participation included Founders Fund, Elad Gil, Alpha Wave, Definition Capital, Positive Sum, Avenir, Vitruvian Partners, Bain Capital Ventures, Ribbit Capital, Atreides, Layer Global.
Source · cognition.com ↗Series C
$400M · $10.2B valuation
After acquiring Windsurf, Cognition said it would continue investing in both Windsurf’s coding editor and its Devin agent.
Led by Founders Fund. Participation included Lux Capital, 8VC, Neo, Elad Gil, Definition Capital, Swish VC, Bain Capital Ventures, Hanabi Capital, D1 Capital.
Source · cognition.com ↗Series B
undisclosed · $4B valuation
Cognition later confirmed that Lux Capital and 8VC jointly led this round. Public reporting gave no exact amount.
Led by Lux Capital, 8VC.
Source · clay.com ↗Funding Round
$175M · $2B valuation
The Information reported a separate financing at a $2 billion valuation after Cognition’s March unveiling of Devin.
Led by Founders Fund.
Source · theinformation.com ↗Seed
$21M
Cognition announced the financing alongside the first public preview of Devin. The launch post called the financing a Series A, while later company round numbering treats it as the seed.
Led by Founders Fund. Participation included Patrick Collison, John Collison, Elad Gil, Sarah Guo, Chris Re, Eric Glyman, Karim Atiyeh, Erik Bernhardsson, Tony Xu, Fred Ehrsam.
Source · cognition.com ↗
Cognition Founding Team
Competitive programming is a shared thread in Cognition’s founding team. Two of its founders won gold medals at the International Olympiad in Informatics before the team built Devin.
- Scott Wu — CEO. Competitive-programming IOI gold medalist.
- Steven Hao — CTO. MIT graduate and IOI gold medalist.
- Walden Yan — CPO. Harvard dropout and competitive programmer.
Cognition Company Development
An Editor Alongside Devin
The Windsurf acquisition added an IDE to Cognition’s autonomous engineering work. It created two points of entry for the company: coding alongside a developer in the editor and taking on assignments through Devin.
Cognition Competitive Landscape
One Repository, Two Styles of Assistance
Windsurf and Cursor compete on the developer’s direct editing workflow. Devin adds a separate mode for work delegated to an agent, so a meaningful comparison distinguishes edits a person supervises in the IDE from assignments intended to run with less direct intervention.
FAQ
Why did a16z — a major Cursor backer — lead a round in Cursor's direct competitor?
Because AI coding stopped looking winner-take-all: a16z led Cognition's US$2B Series E at US$48B, effectively hedging both leaders. Cognition's wedge is the regulated enterprise — FedRAMP High In-Process, a DOE Genesis Mission MOU, Mercedes-Benz's COBOL modernization cut from eight months to eight days — while Cursor holds the developer-desktop base. Thirty-plus firms joined the syndicate.
How did Cognition and Google split Windsurf between them?
Cleanly, by design: roughly a week before Cognition's July 2025 acquisition, Google ran a ~US$2.4B reverse-acquihire taking Windsurf's CEO and ~40 senior R&D staff into DeepMind. Cognition received the product, IP, ~US$82M ARR and ~200 employees — the body without the head — pricing the integrated asset at a US$10.2B Series C two months later: extraction and acquisition, in parallel.
What does a ~53x run-rate multiple price that growth alone doesn't?
Category-infrastructure status: US$48B against nearly US$900M run-rate sits where investors price platforms, not tools — backed by a compounding curve (US$1M in September 2024, US$73M by June 2025, US$492M by May 2026) and enterprise usage up over 10x since the start of 2026. The open question is margin, not demand: compute and server leasing are estimated near US$800M a year.
Why did Cognition raise twice within six weeks of Devin's launch?
The launch financing and the larger investment reported weeks later are separate events. Sources differ on the stage assigned to the launch financing; the current Registry record uses Seed and preserves that conflict. A change in a database label does not establish another financing.
Why does Cognition keep buying small teams instead of building?
Because surfaces, not just models, win enterprises: Windsurf (July 2025) added an IDE and ~US$82M ARR; July 2026 brought TierZero and The Interaction Company's Poke. The tuck-in cadence mirrors Harvey's four 2026 acquisitions ending in Guardrails AI, and World Labs' SceniX deal. Against a US$4–5B annualized target and ~US$800M yearly compute costs, buying teams is the cheaper path to coverage.
Should Devin usage and Windsurf subscriptions be read as the same revenue stream?
They concern different products and may use different billing and reporting measures. Cognition's narrative connects an autonomous coding agent with an IDE acquisition; it does not provide a segment-level revenue reconciliation. Avoid treating a combined run rate as proof of either product's standalone economics.