Cognition Raises $1 Billion at $26 Billion Valuation as Devin AI Engineer Hits $492 Million in Annualized Revenue
The maker of Devin, an autonomous AI software engineer, closed a Series D round led by Lux Capital, General Catalyst, and 8VC, more than doubling its valuation from $10.2 billion eight months earlier.
Overview
Cognition, the startup behind Devin, an autonomous AI software engineer, has raised $1 billion in a Series D round at a $25 billion pre-money valuation — implying a $26 billion post-money figure — according to TechCrunch. The announcement on May 27 marks a valuation that more than doubled from the company’s $10.2 billion post-money figure just eight months earlier, when it closed a $400 million round in September 2025.
What We Know
The round and investors. The Series D was co-led by Lux Capital, General Catalyst, and 8VC, with participation from Ribbit Capital, Atreides, and Layer Global as new investors, and Founders Fund, Elad Gil, Soma Capital, and Omri Casspi among returning backers, per TechCrunch. Cognition has now raised more than $2.5 billion in total, according to The Next Web.
Revenue and growth. The Next Web reports that Cognition’s annualized revenue run rate stood at approximately $37 million in May 2025 and has since grown to $492 million — a roughly 13x increase in twelve months. Enterprise usage has grown 50% month-over-month for the past six consecutive months, according to TechCrunch. Winbuzzer reports that enterprise usage has grown tenfold since the start of 2026, with Devin deployed inside thousands of companies. The company says it aims to cross $1 billion in annualized revenue later in 2026, per The Next Web.
Enterprise customers. Cognition counts Mercedes-Benz, NASA, Goldman Sachs, and Santander among its enterprise customers, according to TechCrunch. Brazilian fintech Nubank is also among its customers, according to Winbuzzer.
Devin’s capabilities and self-use. Devin handles end-to-end software development tasks — planning, writing code, debugging, and deployment — and runs on a mix of Cognition’s proprietary models alongside models from OpenAI and Anthropic, The Next Web reports. CEO and co-founder Scott Wu said that more than 90% of the company’s internal code is now written by Devin, per The Next Web. Devin became generally available for team deployments in December 2024, according to Winbuzzer.
The Windsurf acquisition. In July 2025, Cognition acquired the remaining assets of Windsurf after Google closed a $2.4 billion deal for Windsurf’s talent and licensing, adding to Cognition’s installed customer base, per The Next Web.
What We Don’t Know
- The specific terms or board representation rights associated with the Series D have not been disclosed publicly.
- No pricing details for the new team entry tier mentioned by Winbuzzer have been confirmed.
- The timeline or format for crossing $1 billion in annualized revenue has not been specified beyond later in 2026.
- Cognition has not disclosed details of its proprietary model architecture or how it distributes workloads across its own models versus OpenAI and Anthropic infrastructure.
Analysis
Cognition’s trajectory illustrates the accelerating consolidation of investor capital around a small number of AI coding agent startups that have demonstrated verifiable enterprise revenue. The jump from a $10.2 billion to a $26 billion valuation over eight months — accompanied by a 13x revenue increase — suggests the market is pricing in continued expansion of autonomous software development workflows beyond experimental use.
The competitive backdrop is intensifying. Winbuzzer notes that Anthropic doubled Claude Code’s rate limits and OpenAI released workflow orchestration tools in the same period, indicating that large model providers are moving more aggressively into the agent-led software development market that Cognition helped define. The fact that Cognition itself operates on a mix of proprietary and third-party models — including from its direct competitors — reflects an emerging reality in the AI industry: product differentiation increasingly lives in the orchestration and product layer, not exclusively in the underlying model.