Cognition, Maker of AI Coding Agent Devin, Reportedly in Talks for $40 Billion Valuation
Cognition, the company behind the AI coding agent Devin, is in talks with investors to raise a new funding round at a valuation of at least $40 billion, according to a report published by Bloomberg on August 12, 2026. If the round closes at that figure, it would represent a more than 50% increase over the $26 billion valuation the company secured just three months earlier, when it raised $1 billion in May 2026.
The new valuation is tied to Cognition's revenue trajectory. The company is reportedly on track to reach a $1 billion annualized revenue run rate, according to Bloomberg and corroborated by TechCrunch. An annualized revenue run rate is a way of estimating yearly revenue by taking current monthly revenue and multiplying it by twelve, giving a snapshot of where revenue would land if current conditions held steady.
In May, Cognition's Scott Wu confirmed to TechCrunch that the company had achieved a $492 million annualized revenue run rate, meaning the figure now being cited to investors would represent roughly a doubling over a short period. Wu also said at the time that enterprise usage of Devin was growing 50% month-over-month for the preceding six months.
Cognition has not publicly confirmed the funding talks. Its official blog contains no statement about the reported round or the $40 billion figure as of this capture. Bloomberg broke the story; TechCrunch published its own corroborating report the same day.
The velocity of the valuation climb is notable. Cognition went from $26 billion in May to a reported $40 billion target in August, a span of roughly ninety days. That kind of compression, in my view, reflects investor appetite for AI-native infrastructure plays that can point to actual enterprise revenue rather than research milestones alone. The $40 billion figure places Cognition in rare company among privately held AI companies, and the revenue multiple being implied, even at the $1 billion run rate, is aggressive by any conventional software benchmark.
Devin is positioned as an autonomous coding agent rather than a copilot-style assistant. A copilot suggests code as a developer types, much like autocomplete for software engineering. An autonomous agent, by contrast, takes a task description and works through it independently, making decisions about how to approach the problem and writing the code itself. Wu has told TechCrunch that the product is not being sold as a human replacement; instead, it is frequently deployed on long-tail engineering tasks such as modernizing legacy software or migrating applications between platforms. The company's customer list includes Mercedes-Benz, NASA, and Goldman Sachs.
Cognition has also been building out its market through partnerships and acquisitions. The company partnered with LTM to deploy Devin across LTM's global client base and cybersecurity practice, which serves over 260 clients including 26 of the Fortune 500 and the top five global banks. On its blog, Cognition has announced welcoming The Interaction Company of California, maker of Poke, into the company, as well as TierZero, whose automation work was brought into Devin. Cognition also noted that it and Windsurf combined their work roughly one year ago.
On the government side, Cognition signed a memorandum of understanding with the U.S. Department of Energy to join the Genesis Mission, a national initiative described as "America's Manhattan Project for AI." Devin is FedRAMP Class D (High) In-Process and listed on the FedRAMP Marketplace, positioning it for federal procurement. FedRAMP is the federal government's standardized security assessment process for cloud products; achieving High status means the product meets the government's most stringent security requirements.
Technically, Cognition has been iterating on its own model line. The company launched SWE-1.7, which it described as the most capable model it had trained to date. It also introduced an AI Productivity Guarantee for enterprise customers, committing to fund usage up to $10 million if Devin delivers less engineering value than what was paid for. That guarantee is an unusual bet in a category where return on investment is still being established; it shifts a portion of adoption risk from the buyer to the vendor.
The broader context here is the gap between Cognition's May revenue figure and the $1 billion run rate now being cited to investors. If the company has roughly doubled its annualized revenue in three months, that would be consistent with the 50% month-over-month growth Wu reported, though the compounding math depends heavily on the base month and customer mix. Investors evaluating the $40 billion round will need to assess whether that growth is being driven by net-new enterprise contracts, expanded seat counts within existing accounts, or a combination of both.
The AI coding agent category has attracted significant capital and competitive attention. What distinguishes Cognition's trajectory is less the existence of the category than the revenue numbers attached to it. A $1 billion run rate at a private company less than two years after Devin's public debut, if accurate, would place Cognition among the fastest-growing enterprise software companies on record by that metric. Whether the $40 billion valuation is justified by current revenue or reflects a bet on the addressable market for autonomous coding agents is the question investors are now weighing.
