Cognition Eyes $40 Billion Valuation Three Months After Billion-Dollar Round
The maker of Devin, an AI coding agent handling enterprise grunt work, is negotiating its next raise on the back of accelerating revenue growth and expanding Fortune 500 adoption.
A Compressed Timeline for Unicorn Acceleration
Cognition, the San Francisco startup behind Devin, is in preliminary discussions with investors for a fresh funding round that would value the company at a minimum of $40 billion. The talks come barely three months after the firm closed a $1 billion raise at a $26 billion valuation in May, underscoring the velocity at which capital is chasing AI infrastructure plays that demonstrate enterprise traction.
The proposed valuation hinges on Cognition reaching a $1 billion annualized revenue run rate, a threshold that appears within striking distance given the company's recent trajectory. In May, founder Scott Wu disclosed that Cognition had already achieved $492 million in annualized recurring revenue, with enterprise customers expanding their Devin usage by 50 percent month-over-month for the preceding six months. If that pace held through the summer, the billion-dollar milestone would be well in sight by the fourth quarter.
At DailyTechWire, we've tracked dozens of AI infrastructure raises across the region this year, and Cognition's tempo stands out even in a frothy market. The gap between a $26 billion and $40 billion valuation in a single quarter reflects investor conviction that coding agents represent a category with durable unit economics, not just a ChatGPT wrapper with enterprise branding.
What Devin Actually Does
Devin is not a replacement for software engineers. Wu has been emphatic on this point, and the product's design bears it out. The agent specializes in what Wu calls "long-tail grunt-work": migrating legacy applications from one platform to another, updating deprecated dependencies, refactoring codebases to meet new compliance standards, and other high-volume, low-creativity tasks that human engineers find tedious but that enterprises must complete to avoid technical debt.
This positioning matters. The AI coding market is crowded with GitHub Copilot, Cursor, and a swarm of startups pitching autocomplete on steroids. Cognition carved out a niche by targeting work that sits outside the inner loop of feature development, work that enterprises are willing to pay for at scale because the alternative is burning senior engineering time or letting systems ossify.
The customer roster supports the thesis. Cognition lists Mercedes-Benz, NASA, and Goldman Sachs among its users. These are organizations with massive installed bases of legacy software, strict migration timelines, and budget authority to pay for tools that accelerate compliance and modernization without adding headcount. NASA, for instance, operates mission-critical systems written in languages and frameworks that predate the smartphone era; an agent that can semi-automate the update cycle has obvious appeal.
The Unit Economics Argument
Why would investors stretch to a $40 billion valuation for a company that, even at the higher revenue threshold, would still be trading at 40 times annualized revenue? The bet rests on two assumptions. First, that Devin's seat-based or usage-based pricing can scale without a corresponding increase in customer acquisition cost. Enterprise software with strong product-led growth dynamics, especially in infrastructure categories, can sustain high multiples if net revenue retention exceeds 130 percent. Cognition's 50 percent monthly expansion rate among existing customers suggests it is hitting that bar.
Second, that the addressable market for code modernization and maintenance is far larger than the market for net-new feature development. Every Fortune 500 company runs software that needs constant care; most of it is invisible to end users but critical to operations. If Devin can capture even a fraction of the budget currently allocated to offshore development shops and internal platform teams, the revenue ceiling is measured in tens of billions, not hundreds of millions.
There is risk, of course. OpenAI, Anthropic, and Google are all investing heavily in coding models, and any of them could bundle a Devin-like agent into their enterprise offerings at marginal cost. Cognition's moat depends on workflow integration, fine-tuning for specific enterprise environments, and the trust that comes from being a specialist rather than a platform add-on. That moat is real but not infinitely wide.
The Broader Pattern in AI Funding
Cognition's trajectory mirrors a pattern we've observed across Asia and North America: startups that demonstrate $100 million-plus ARR and triple-digit net retention are commanding valuations that would have seemed absurd in the 2021 bubble but are now table stakes. Lovable, another AI-native developer tool, recently confirmed a $13.3 billion valuation. River AI, barely two months old, raised $1.1 billion from General Catalyst. The common thread is evidence of enterprise spending, not just pilot programs.
For Cognition, the $40 billion target is a forcing function. It signals to the market that the company expects to cross $1 billion in ARR by year-end and that it plans to use the capital to expand beyond coding into adjacent automation categories. Wu has hinted at ambitions in infrastructure orchestration and security remediation, both of which share Devin's DNA: high-volume, low-glamour work that enterprises will pay to offload.
Whether Cognition closes the round at that valuation depends on execution over the next quarter and whether the broader venture market remains open to writing nine-figure checks. But the fact that the conversation is happening at all, three months after a billion-dollar raise, tells you where the center of gravity in AI investment has shifted: toward companies that are already generating revenue at scale, in markets where the alternative to software is hiring humans.
