August 15, 2026
TechCrunch reported on August 12 that Cognition, the company behind AI coding agent Devin, is in early talks to raise new funding at a $40 billion valuation. That's up from $26 billion just three months ago, when Cognition closed a $1 billion round in May 2026. The jump is backed by real growth: Cognition reportedly hit a $1 billion annualized revenue run rate, up from $492 million as of that May raise, after six straight months of 50% month-over-month growth. Customers cited include Mercedes-Benz, NASA, and Goldman Sachs.
Numbers like that don't happen in a category nobody wants. Whatever else is true about AI hype cycles, this is a real signal that AI-assisted software development has become a business worth tens of billions of dollars in enterprise value, not a demo. We take that seriously. It's good news for anyone who believes AI is going to change how software gets built and who gets to build it.
The part of the story that matters most to us isn't the valuation -- it's how Cognition's own founder and CEO, Scott Wu, described what Devin does. According to the Bloomberg reporting TechCrunch cited, Wu said Devin 'is not being sold as a human replacement,' but instead handles 'long-tail grunt-work that many programmers dislike' -- things like legacy software updates and platform migrations.
That is a precise and honest description of the product, and it tells you exactly who Devin is for. It's for programmers. It sits inside an existing codebase, an existing review process, an existing engineering workflow -- and it takes the tedious 20% of that workflow off a human's plate so the humans can focus on the interesting 80%. Mercedes-Benz, NASA, and Goldman Sachs don't buy Devin because they lack engineers. They buy it because they have large engineering organizations with enormous backlogs of unglamorous work, and Devin makes those organizations faster at the job they're already doing.
Here's the thing Wu's framing makes obvious once you sit with it: Devin assumes an engineering org already exists. Someone has to define the codebase, run the review, own the architecture, and decide what 'done' means. Devin accelerates that team. It does not replace the need for one, and it was never built to.
Most small businesses, ops teams, and non-technical founders don't have that team and don't want to build one. Their problem isn't 'our engineers are too slow at grunt work.' Their problem is 'we don't have engineers, and hiring a team just to get an internal tool or a client-facing app is a six-figure detour we can't afford.' That's not a smaller version of Cognition's market -- it's a genuinely different buyer with a genuinely different job to be done. We've written before about why the buy vs. build vs. ViibeStack decision usually has nothing to do with engineering velocity and everything to do with whether engineering exists in the building at all.
ViibeStack's AI App Builder starts from a plain-language description of what someone needs -- a client portal, an equipment tracker, an approval workflow -- and produces a working app without requiring the person asking for it to know what a pull request is. There's no codebase to onboard an agent into, because the whole point is that the business owner never had to build or hire for one in the first place. Compare that to how we describe our own how it works flow: the person describing the app is the same person who ends up using it, with nobody translating requirements into engineering tickets in between.
That's why the AI-coding funding boom -- Devin and the wave of pro-developer agents chasing it -- and the AI-app-building category we're in are adjacent, not competitive. They both got here by making AI write software, but they're solving for opposite starting conditions. One assumes you already have a dev team you want to make faster. The other assumes you don't, and shouldn't need one. We've made a version of this same argument before about ViibeStack vs. Microsoft Power Apps and the broader AI coding bubble question -- the tools that win are the ones honest about who they're actually built for.
If you already have an engineering team, a tool like Devin is a reasonable bet: it can chew through migrations and legacy maintenance so your engineers spend more time on the work that actually needs a human. Cognition's numbers suggest plenty of large organizations agree.
If you don't have an engineering team -- and don't want the overhead of hiring, managing, and retaining one just to get a working internal tool or app -- that's not a smaller version of the same problem. It's a different problem entirely, and it's the one we built ViibeStack to solve. Browse our templates or read a real example in our Famanager customer story to see what that looks like when there's no engineering org involved at all.
Sources