← All posts / Industry

Cognition AI Eyes $40 Billion Valuation as Devin Nears $1 Billion ARR

The maker of autonomous coding agent Devin is in early funding talks that could value it at $40 billion or more, with annualized revenue approaching the $1 billion mark.

Cognition AI Eyes $40 Billion Valuation as Devin Nears $1 Billion ARR

Just three months after raising $1 billion at a $26 billion valuation, Cognition AI is already back at the negotiating table. According to a Bloomberg report published August 12, 2026, the San Francisco–based startup behind the autonomous coding agent Devin is in early discussions with investors for a new round that could push its valuation past $40 billion — a jump of more than 50 percent in a single quarter.

From $37 Million to Nearly $1 Billion in a Year

The valuation chatter is anchored in a revenue trajectory that has left even seasoned AI investors doing a double take. Cognition’s annualized revenue run rate sat at roughly $37 million in May 2025. By May 2026, when the company closed its $1 billion Series D led by Lux Capital, General Catalyst, and 8VC, that figure had climbed to $492 million — a 13-fold increase in twelve months. Now, according to Silicon Republic’s reporting on the Bloomberg scoop, the company’s annualized run rate is nearing the $1 billion mark, a threshold that would place it among the fastest-growing software businesses in history.

That growth has been driven primarily by self-serve adoption of Devin, the autonomous coding agent that Cognition launched in March 2024 as “the world’s first AI software engineer.” Unlike code-completion tools such as GitHub Copilot or inline assistants like Cursor, Devin operates as a cloud-based agent that can take a natural-language task description, spin up a sandbox environment, write and test code across multiple files, and submit a pull request — all with minimal human intervention. Cognition’s CEO Scott Wu told investors that Devin now writes approximately 95 percent of the company’s own code, a claim that underscores how far autonomous coding has advanced in two years.

The Windsurf Acquisition and the Path to $40 Billion

Cognition’s rapid ascent has been accelerated by a series of bold strategic moves. In July 2025, the company acquired Windsurf, a competing AI-powered IDE startup, for an estimated $3 billion. The deal came just days after Google DeepMind poached Windsurf’s CEO and research leadership, and followed a failed acquisition bid by OpenAI. Cognition folded Windsurf’s 350 enterprise customers and hundreds of thousands of daily active users into its platform, rebranding the IDE as “Devin Desktop” in June 2026.

The Windsurf acquisition more than doubled Cognition’s revenue base overnight, according to reporting from thenextweb.com and Sacra. Enterprise contracts with Goldman Sachs, Mercedes-Benz, and even the U.S. Army have added high-margin, multi-year revenue that has proved stickier than the self-serve subscriptions that powered Devin’s early growth. With the integrated Windsurf user base and aggressive enterprise expansion, monthly revenue growth has averaged 50 percent for six consecutive months.

The new funding round, still in its early stages, would provide capital for Cognition to scale its infrastructure, expand its model-training capabilities, and fend off a growing field of competitors. Bloomberg’s sources indicated that the lead investors and round size have not yet been finalized, and that the $40 billion figure represents a floor rather than a ceiling — the final valuation could go higher depending on demand.

A Crowded and Cutthroat Market

Cognition’s potential $40 billion valuation lands in the middle of one of the most fiercely contested categories in AI. Autonomous and semi-autonomous coding tools have exploded since Devin’s debut, with multiple well-capitalized players vying for developer mindshare:

  • Cursor (Anysphere), the AI-native code editor, has raised over $900 million and was valued at $9.9 billion in its most recent round.
  • Claude Code, Anthropic’s terminal-native CLI agent built on the Claude Opus 5 model, has become the top-rated agentic coding system in 2026 reviews, with a 72.6 percent code acceptance rate in a peer-reviewed arXiv study.
  • OpenAI Codex leads the field in code acceptance rate at 79.9 percent according to the same study, and has surpassed 104,000 GitHub stars.
  • opencode, an open-source agent, has become the most-starred coding tool on GitHub with over 193,000 stars.

What sets Cognition apart is its head start in fully autonomous, end-to-end task execution. While Cursor and Claude Code excel at augmenting developers within their existing workflow — suggesting completions, refactoring code, answering questions — Devin is designed to operate independently, taking ownership of entire engineering tasks from ticket to merged pull request. The company has positioned this as a fundamentally different value proposition: not a productivity multiplier for existing engineers, but a scalable engineering workforce that can be deployed in parallel.

That positioning has resonated with enterprises facing chronic developer shortages. In a July 2026 interview, Scott Wu noted that Cognition’s largest customers now run dozens of Devin agents simultaneously, each handling separate tasks across a codebase — a pattern that maps more closely to managing a team of offshore contractors than to using a development tool.

Why Investors Are Willing to Pay Up

The $40 billion valuation, if it materializes, would represent roughly 40 times Cognition’s projected forward revenue run rate of $1 billion — a multiple that seems astronomical by traditional software standards but is consistent with how venture investors are pricing the AI infrastructure and agentic AI categories. For comparison, OpenAI was valued at $730 billion in February 2026 at roughly 25 times revenue, and Anthropic’s most recent private valuation implied a similar multiple on its reported $30 billion ARR.

The premium for Cognition reflects three factors. First, the agentic coding market is widely seen as the most immediately monetizable application of large language models — developers are willing to pay, the ROI is measurable in hours saved, and adoption has been remarkably frictionless. Second, Cognition’s revenue growth rate (50 percent month-over-month for six months) is among the fastest ever recorded in enterprise software, exceeding even the early trajectories of Slack, Zoom, or Snowflake. Third, the network effects of Devin’s codebase — the agent learns from every task it completes — create a defensible moat that compounds with scale.

Risks and Open Questions

Not everyone is convinced. Critics point to Cognition’s heavy burn rate, the commoditization pressure from open-source alternatives like opencode, and the reality that foundation model providers (OpenAI, Anthropic, Google) could integrate autonomous coding capabilities directly into their platforms at any time, undercutting standalone players. The competitive landscape has also shifted rapidly: Devin’s code acceptance rate of 68.0 percent in the February 2026 arXiv benchmark study lagged behind OpenAI Codex (79.9 percent), Cursor (74.4 percent), and Claude Code (72.6 percent), raising questions about whether Cognition’s technology lead has eroded as larger players have caught up.

There is also the question of whether the current funding environment is sustainable. Q1 2026 saw $300 billion in global venture funding poured into 6,000 startups, a 150 percent increase quarter-over-quarter, with OpenAI, Anthropic, xAI, and Waymo alone absorbing roughly 65 percent of every venture dollar deployed worldwide. If the AI capital cycle cools, companies burning cash to maintain hypergrowth — even those with $1 billion in revenue — could face sharp valuation corrections.

Still, for now, Cognition’s trajectory is a testament to how quickly the AI coding agent category has matured. What was a novelty demo in March 2024 — an AI that could independently build and deploy a web application from a prompt — is now a billion-dollar business with the U.S. military, global banks, and automakers as paying customers. Whether a $40 billion valuation proves justified will depend on whether Devin can maintain its growth pace in an increasingly crowded field, and whether the autonomous coding paradigm it pioneered becomes the default way software is built.