Cognition raises $2 billion as AI coding startup hits $48 billion valuation

AI coding startup Cognition has raised more than $2 billion in a Series E funding round, pushing its valuation to approximately $48 billion as investor interest in AI-powered software development continues to accelerate.

The latest funding comes less than four months after the company announced a valuation of $26 billion, highlighting the rapid increase in investor confidence in its technology.

Cognition was founded in 2024 by Scott Wu and is best known for developing Devin, an AI-powered coding assistant designed to perform software engineering tasks with greater autonomy.

The Series E round was led by new investors Andreessen Horowitz and Accel, with existing backers Founders Fund, General Catalyst and Avenir also participating.

Other investors in the round included Benchmark, Bessemer Venture Partners, Kleiner Perkins, Greylock and Lightspeed.

Cognition said its annualized revenue run rate has increased sharply, rising from $492 million in May to nearly $900 million currently.

The company did not disclose the precise methodology used to calculate the figure, although annualized revenue run rates are commonly estimated by multiplying a representative month’s revenue by 12.

Devin is already being used by major organizations across several industries, according to Cognition.

Its customers include Nvidia for chip-design work, GE Aerospace for aviation-related projects, Citigroup for financial services, Mercedes-Benz Group for automotive engineering and Modal Labs for AI infrastructure.

Beyond traditional coding assistance, Devin is designed to function as an autonomous AI agent capable of investigating incidents, identifying security vulnerabilities and prioritizing issues.

The system can also perform tasks triggered by activity across tools such as Slack, GitHub and Linear without requiring users to manually initiate every interaction through a chat interface.

Cognition has expanded its operations internationally, with offices in cities including Washington, D.C., Tokyo, Singapore, London, São Paulo, Madrid, New York and Austin.

The company is also facing substantial infrastructure expenses as it scales its AI operations.

Cognition leases a large Nvidia server cluster that costs millions of dollars each year, while its overall cash burn could reach approximately $800 million in 2026.

To gain greater control over its technology and reduce reliance on external AI providers, Cognition is developing its own AI models using open-source alternatives.

The company says its broader objective is to address the growing demand for software while overcoming the limited capacity of human engineering teams.

Cognition envisions software engineers taking on more of an architectural role while delegating portions of execution to groups of AI agents.

The company is also developing what it calls an independent agent laboratory, allowing it to combine different AI models depending on the requirements of a particular task.

That strategy could allow Cognition to use its own models alongside third-party and open-source systems rather than relying exclusively on a single AI provider.

Competition in the AI coding sector has intensified as startups race to build increasingly capable software-development agents.

Cursor, another major player in the market, was reportedly discussing a funding round at a valuation near $50 billion before reaching an agreement to be acquired by SpaceX for $60 billion.

Cursor had reportedly surpassed $2 billion in annualized revenue around the time of those funding discussions, underscoring the scale of competition in the sector.

Cognition is reportedly targeting annual revenue of between $4 billion and $5 billion by the end of 2026, while Cursor is expected to reach around $6 billion.

Andreessen Horowitz’s investment in both companies also suggests that the venture firm sees substantial room for multiple major competitors in the rapidly expanding AI coding market.