AI coding is heating up again. Cognition, the startup behind the AI assistant Devin, has closed a $2 billion funding round, now boasting a $48 billion valuation. That’s up sharply from just $26 billion just four months ago—and it suggests investors believe there’s room for multiple winners in the AI coding space.
The round was led by names like Andreessen Horowitz, Accel, Founders Fund, General Catalyst, and Avenir. Cognition also revealed that its annualized run-rate revenue jumped from $492 million in May to $900 million in the latest period. Though the startup didn’t break down exactly how it calculates that, the number implies monthly revenue multiplied by 12.
How Cognition stacks up against Cursor
Cursor, another major name in AI coding assistants, had been in talks earlier this year to raise capital at around a $50 billion valuation. It later sold to SpaceX for roughly $60 billion. At that time, Cursor’s yearly revenue run-rate had surpassed $2 billion—making Cognition’s valuation multiple noticeably higher despite its lower revenue base. Cursor’s sale was largely driven by compute constraints that investors said limited its scale.
Cognition is trying to avoid that pitfall. It leases an NVIDIA server cluster that comes with hundreds of millions in annual costs. Some estimates put its cash burn as high as $800 million this year. To offset that, the company is building more of its own model infrastructure rather than relying on third-party models from players like OpenAI and Anthropic. That move could help reduce costs and move Cognition toward profitability over time.
For perspective, Cognition is expected to end 2026 with $4–5 billion in annualized revenue. Cursor, in contrast, had been projected by spring to surpass $6 billion by year-end. Despite the similarities, this valuation round confirms investors aren’t banking on one dominant giant in AI coding—they’re spreading bets.
Founded in 2024 by math prodigy Scott Wu, Cognition already counts enterprise giants like Mercedes-Benz, NASA, Goldman Sachs, and Citi among its customers. Even though it’s burning through cash fast, its rapid revenue growth, big clients, and move toward proprietary model infrastructure are making investors confident that its path might be sustainable.
It’s a different narrative than the “winner-takes-all” story that’s often been told about AI. Cognition’s big valuation and aggressive investment in its infrastructure indicate that early market signals point to something more fragmented—and possibly more competitive.
Why this matters: AI coding tools are increasingly central across industries. General-purpose AI models from OpenAI, Anthropic, and others are powerful but expensive at scale. Companies that train their own models—if done correctly—have the potential to reduce dependency and cost. Because computer infrastructure and energy costs are high, those that lock in efficient pipelines, fresh training data, and scalable models stand to succeed. Investors putting huge money behind Cognition now could reshape who wins in AI developer tooling.
