> ## Content Index
> Fetch the complete content index at: https://www.wwconemedia.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Cognition Just Hit a $48 Billion Valuation as Devin Nears $900 Million in Run-Rate Revenue — But One Number Shows How Expensive the AI Race Could Become
- URL: https://www.wwconemedia.com/cognition-just-hit-a-48-billion-valuation-as-devin-nears-900-million-in-run-rate-revenue-but-one-number-shows-how-expensive-the-ai-race-could-become/
- Published: 2026-09-09T01:29:10.000Z
- Updated: 2026-09-09T01:29:10.000Z
- Author: WWC NEWSDESK
- Tags: ASIA, BUSINESS

**SAN FRANCISCO, Sept. 9, 2026** — Cognition, the artificial intelligence startup behind the Devin autonomous software-engineering agent, has raised **more than $2 billion in Series E funding at a $48 billion valuation**, marking another extraordinary jump for a company founded only in 2024.

The round was led by new investors **Andreessen Horowitz and Accel**, while existing backers including **Founders Fund, General Catalyst and Avenir** also participated. Cognition disclosed a much broader roster of investors that includes Benchmark, Bessemer Venture Partners, Kleiner Perkins, Greylock, Lightspeed, T. Rowe Price, Lux Capital, Bain Capital Ventures and Nvidia.

The financing comes only months after Cognition raised more than $1 billion at a **$26 billion post-money valuation** in May, meaning investors have placed a dramatically higher price on the business in a remarkably short period.

But the number attracting just as much attention as the $48 billion valuation is Cognition's revenue trajectory.

## Revenue Has Jumped From $492 Million to Almost $900 Million Run Rate

Cognition said its **run-rate revenue has increased from $492 million in May to almost $900 million today**.

That is an important distinction from ordinary annual revenue.

Run-rate revenue generally annualizes the company's current pace of sales. It therefore gives investors an indication of where revenue could land if present business conditions continue, but it should not be interpreted as Cognition having already recorded $900 million during a completed 12-month financial year.

Still, the pace of expansion helps explain why investors have been willing to dramatically increase Cognition's valuation.

The Information reported that Cognition is currently generating roughly **$75 million per month**, equivalent to about $900 million on an annualized basis. It also reported that the company's rapid expansion comes with substantial computing costs and that cash burn could reach about $800 million this year. That cash-burn estimate has not been confirmed in Cognition's public funding announcement.

That tension — enormous growth coupled with enormous computing requirements — could become one of the most important tests facing the current generation of AI startups.

## Devin Is Moving Beyond Being Just an AI Coding Assistant

Cognition's main product, **Devin**, is designed to handle software-engineering work with far greater autonomy than a traditional code-completion tool.

Rather than merely suggesting the next line of code, autonomous coding agents are designed to receive an objective, plan the work, navigate a codebase, write and modify software, test changes and complete larger engineering tasks.

Cognition says Devin is now being deployed across major companies including **Nvidia for chip design, GE Aerospace in aviation, Citi in financial services, Mercedes-Benz in automotive and Modal in AI infrastructure**.

The company is also extending Devin deeper into engineering operations.

Among the features Cognition highlighted are **Auto-Triage**, which can conduct an initial investigation of software incidents; **Security Swarm**, which is designed to locate and triage vulnerabilities; and **Automations**, which can begin work when events occur in platforms such as Slack, GitHub and Linear.

That matters because the long-term prize in AI coding may not simply be helping developers type code faster.

The much larger opportunity is software that can increasingly **perform entire engineering workflows with limited human intervention**.

## Cognition's Bigger Vision: Engineers Become Architects

Cognition is explicitly betting on what it calls the emerging era of **“self-driving software.”**

Its vision is that human software engineers increasingly define objectives, architecture and priorities while groups of AI agents handle more execution.

The company says engineering teams could eventually delegate work to “swarms” of agents rather than manually performing every individual development task.

That is a much more ambitious proposition than autocomplete.

If the model works, AI coding agents could potentially address one of technology's persistent constraints: companies routinely have more software projects, upgrades, migrations, security fixes and maintenance work than their engineering teams have time to complete.

Cognition argues AI agents could dramatically expand that engineering capacity.

But whether enterprises will entrust increasingly critical software systems to autonomous agents — and how much human supervision those agents will continue to require — remains one of the central unanswered questions in the sector.

## Cognition Is Expanding Globally — Including Singapore

The company is no longer operating as a small Silicon Valley-focused startup.

Cognition says it has opened offices in **Washington, Tokyo, Singapore, London, São Paulo and Madrid** during the past year, alongside hubs in San Francisco, New York and Austin.

The Singapore expansion is particularly noteworthy for Southeast Asia as global AI companies increasingly establish engineering, enterprise-sales and regional operations in the city-state.

Cognition says its strategy is also to remain an **independent AI-agent company** rather than depend on a single underlying model provider.

That allows it, in principle, to choose among different AI models — including its own technology — according to the task being performed.

## The AI Coding War Is Becoming One of Tech's Most Expensive Battles

Cognition is not operating alone.

AI-powered software development has become one of the most fiercely contested areas of artificial intelligence, drawing competition from both specialist startups and major foundation-model developers.

TechCrunch noted earlier this year that Cognition faces an increasingly crowded field that includes **Anthropic's Claude Code, OpenAI's Codex and Google's coding tools**.

The competitive landscape became even more dramatic in 2026.

TechCrunch reported that SpaceX completed a roughly **$60 billion acquisition of Cursor**, another major AI coding company, while Cognition CEO Scott Wu publicly disputed a separate report that SpaceX had attempted to acquire Cognition, saying his company was not for sale.

That leaves Cognition positioned as one of the largest independent companies still competing for the emerging autonomous-software-development market.

And investors appear convinced that the sector may support more than one enormous winner.

TechCrunch described Cognition's latest valuation as evidence that investors still see AI coding as something other than a simple winner-take-all market.

## Windsurf Gave Cognition Another Weapon

Cognition's rapid expansion also follows its 2025 acquisition of the remaining business of **Windsurf**, the AI coding platform.

That transaction came after Google secured Windsurf's CEO, co-founder and several senior researchers through a separate $2.4 billion talent-and-licensing arrangement.

Cognition's deal included Windsurf's intellectual property, product, trademark, brand and remaining team, giving Cognition another significant foothold in the AI developer-tools market.

At the time, Cognition said it planned to invest heavily in integrating Windsurf's technology into its products.

That acquisition helped turn the company from essentially the maker of Devin into a broader AI software-engineering platform.

## Investors Are Still Pouring Billions Into AI

Cognition's financing is also part of a much wider capital race.

On the same day Cognition announced its Series E, French AI company **Mistral AI** announced a multibillion-euro financing that valued it at roughly $24 billion, highlighting the scale of capital still flowing toward companies that investors believe could become major AI platforms.

But investors are also becoming more selective.

AI companies increasingly need to demonstrate not simply impressive models or product demonstrations but paying customers, repeat usage and credible enterprise adoption.

That is where Cognition's nearly $900 million revenue run rate becomes especially important.

The company's valuation may look extraordinary for a business only a few years old, but investors are effectively wagering that autonomous software agents could become a major layer of the global software industry.

## The $48 Billion Question

There is still a major difference between proving that companies will experiment with AI coding agents and proving that those agents can permanently capture tens of billions of dollars in economic value.

Cognition must continue expanding revenue while paying for the considerable computing infrastructure required to run increasingly capable agents.

It must also compete against some of the best-funded companies in technology.

And its products will have to demonstrate that autonomous agents can reliably operate inside complicated corporate software environments without introducing unacceptable security, reliability or governance risks.

Yet the latest funding round shows investors are willing to place a very large bet on the outcome.

In May, Cognition was valued at $26 billion.

Four months later, it says its revenue run rate is approaching $900 million and investors are valuing the company at **$48 billion**.

The bigger question now is no longer whether AI can help programmers write code.

It is whether companies such as Cognition can convince businesses that **AI should increasingly be trusted to do the engineering work itself**.

And if that transition happens at scale, today's $48 billion valuation may turn out to be less about Devin's current revenue than investors' belief that the entire job of building software is about to change.

WWC ONE MEDIA M.J.E