Cognition Tripled Its Valuation in 14 Weeks. The Revenue Multiple Didn't Move.
We build agentic products, so Cognition’s fundraising trajectory isn’t abstract market news to us — it’s a read on how the market is pricing the exact category we work in. Cognition, the company behind the AI coding agent Devin, is in the final stages of a round that values it at roughly $47–48 billion, up from $26 billion in May 2026 and $10.2 billion a year before that. That’s three valuation step-ups in thirteen months, and the most recent one happened in fourteen weeks.
The numbers
Bloomberg first reported the raise on September 2. By September 8, Yahoo Finance’s reporting had the round grown past $2 billion, led by new investors Andreessen Horowitz and Accel, with returning backers Founders Fund, General Catalyst, and Avenir.
The funding timeline, in full:
- August 2025: $500M at a $9.8B valuation
- September 2025: $400M at a $10.2B valuation (Founders Fund-led)
- May 2026: $1B+ at a $26B valuation
- September 2026: $2B+ at a $47–48B valuation (a16z, Accel lead)
Founded in late 2023 by Scott Wu, Steven Hao, and Walden Yan, Cognition’s annualized recurring revenue nearly doubled between the last two rounds — from $492 million in May to roughly $900 million by September, with an internal target of $4–5 billion by year-end. The detail worth sitting with: the valuation-to-revenue multiple stayed almost exactly constant, around 53x, across both rounds. Investors aren’t bidding the multiple up — they’re betting the hypergrowth rate holds.
Enterprise customers cited in reporting include Nvidia, GE Aerospace, Citi, Mercedes-Benz, Goldman Sachs, Dell, Santander, the U.S. Army, and the U.S. Navy. The anecdote doing the most work in Cognition’s pitch: Devin reportedly completed a Mercedes-Benz legacy-system migration in 8 days that was internally estimated to take 8 months. Cognition also says Devin now writes 89% of Cognition’s own code.
The part that complicates the clean story
Nvidia is both an investor in and a customer of Cognition. Multiple outlets flagged that dual role as muddying how cleanly you can read the revenue. It’s the same circularity pattern — a company’s own investor becoming its revenue — that shows up across AI infrastructure financing more broadly right now, and it’s a specific, named instance of a concern that’s usually discussed in the abstract.
The other number worth holding next to the $900M ARR: Cognition’s annual cash burn may reach roughly $800 million, largely on infrastructure costs. Growing revenue and losing nearly a dollar for every dollar of ARR aren’t contradictory at this stage of a company — but it’s the specific tradeoff being priced at $47 billion, not a footnote to it.
It’s not just Cognition
The round makes more sense read against the rest of the category. Cursor’s parent company, Anysphere, was acquired by SpaceX for $60 billion — a deal announced in June and closed August 14, folding Cursor into a new “SpaceXAI” unit. A rocket and satellite company buying a coding-tool company for more than Cognition’s entire current valuation is, on its own, a stranger data point than anything in Cognition’s own round. Lovable raised $400M at a $13.3B valuation in August. Replit raised $400M at a $9B valuation in March. Anthropic’s Claude Code and OpenAI’s Codex-style tooling compete in the same lane from inside the frontier labs themselves. Nearly every well-capitalized player in AI, general-purpose or not, is now trying to own “AI writes your code.”
Why it matters to us specifically
Coding is one of the few agentic-AI use cases with provable, measurable ROI — the Mercedes anecdote is exactly the kind of story that closes enterprise deals, and revenue at Cognition is growing roughly in line with valuation rather than wildly behind it. That’s the strongest version of the bull case, and it’s real. Set against it: a $53x-revenue-multiple business burning $800M a year to sustain that growth, with its own largest investor also showing up as a customer, is precisely the profile that bubble-skeptics point to when they ask whether AI infrastructure spending is fundamentally sound or just well-marketed. Both of those things are true about the same company at the same time. We don’t think there’s a way to average them into a single verdict that means anything.
What we’re watching
Whether the round closes at the reported size — some reporting suggested investor demand could push it toward $10B. Whether Cognition actually hits $4–5B ARR by year-end, a jump that would need to roughly quintuple current revenue in one quarter. Whether SpaceX’s Cursor integration produces a genuinely different product or turns out to be a talent and IP acquisition wearing a product story. And whether Anthropic or OpenAI answer this round of consolidation with a major coding-agent move of their own — which would tell us more about where this category is actually headed than any single valuation number does.