What happened
Cognition, the startup behind the autonomous coding agent Devin, raised more than $1 billion on May 27 at a $25 billion pre-money valuation, putting the company's post-money value at roughly $26 billion. The round was led by Lux Capital, General Catalyst and 8VC, with existing backers including Founders Fund also participating.
CEO Scott Wu told TechCrunch and Bloomberg Television that the raise was structured to keep Cognition independent rather than fold into a larger AI lab. The number behind the valuation jump is the more interesting part: Cognition disclosed an annualized revenue run rate of $492 million, up from about $37 million a year earlier. That is a roughly 13-fold increase in twelve months, and the company said enterprise usage of Devin has grown 50% month over month for six consecutive months.
Why the growth curve looks different from the rest of AI funding
Most of the AI capital raised this year has gone toward inference and infrastructure: chips, data centers and the pipes that move data between models. Cognition's round is a rare data point on the other side of that spending, evidence that a specific enterprise application, in this case autonomous code generation and maintenance, is generating recurring revenue at scale rather than usage in pilot programs that never convert.
Cognition's customer list has expanded well past early-adopter developer tools. The company has named Goldman Sachs, Citi, Dell, Cisco, Palantir, Nubank and Mercado Libre among organizations using Devin, alongside customers gained through its 2025 acquisition of the AI coding IDE Windsurf. That deal added roughly $82 million in ARR and more than 350 enterprise customers with under 5% overlap with Cognition's existing base, according to figures the company disclosed at the time. Combining a developer-facing product with an IDE that engineering teams already had open all day appears to have been the mechanism behind the jump in usage, not a single flagship contract.
What this means for enterprise engineering teams
For companies evaluating AI coding tools, the relevant signal is not the valuation number but the usage pattern underneath it. A 50% month-over-month growth rate sustained for half a year suggests engineering organizations are expanding Devin's scope inside existing accounts, not simply adding new logos. That distinction matters for any CTO deciding whether to budget for an autonomous coding agent as a line item rather than a pilot: sustained expansion inside accounts that already pay is a harder number to manufacture than a headline customer logo.
It also raises the stakes for competing coding agents from GitHub, Anysphere's Cursor and other vendors, all competing for the same engineering budgets. Cognition's pitch is that Devin operates with less human review than a copilot-style tool, taking on tasks end to end rather than suggesting code inline. Whether that approach holds up as usage scales to organizations with stricter code-review requirements than Cognition's current customer base is the open question the company has not yet had to answer publicly.
The funding context
At $26 billion, Cognition's valuation now sits well above where enterprise software companies with comparable revenue have historically traded, reflecting investors pricing in continued triple-digit growth rather than current revenue alone. The company has raised aggressively: it closed a $400 million round in September 2025 at a $10.2 billion valuation, meaning its valuation has increased roughly two and a half times in eight months. Reports have already surfaced that Cognition is in early talks for a subsequent round that would value the company near $40 billion, tied to reaching a $1 billion annualized run rate, though no terms have been finalized as of this writing.
That pace is unusual even by the standards of this funding cycle, and it puts pressure on Cognition to keep the growth rate from its disclosed numbers intact. A company raising this quickly against revenue that is still concentrated in a handful of large accounts has less room to absorb a slowdown without the next round becoming a difficult conversation.
The bottom line
Cognition's raise is a funding story, but the number that matters is the revenue behind it. Enterprises are now paying for an AI agent to write and maintain code, at scale, inside organizations with the kind of compliance and audit requirements that make pilot theater difficult to sustain. Whether that holds as the customer base broadens past early adopters like Goldman Sachs and Palantir into companies with less appetite for autonomous code changes is the question the next twelve months will answer.
Sources: TechCrunch · Bloomberg · The AI Insider · Winbuzzer