The myth that AI is replacing engineers, in three numbers from this week
Cognition just hit $25 billion selling Devin as the AI that replaces engineers. IBM rolled out its own coding agent to 80,000 engineers the same week. Here is what those two numbers, plus one from Microsoft, actually say about who is running the code.
TLDR This week Cognition hit a $25 billion valuation selling Devin as the AI that replaces engineers, while IBM rolled out its own internal coding agent to 80,000 employees who are still very much employed. The replacement narrative sells. The deployment data tells a different story, and the gap is the thing executives need to budget for in Q2. The myth I have been counting the times I heard “AI is replacing engineers” this week, and I gave up after twelve. The latest round started Sunday when news broke that Cognition was raising at a $25 billion valuation. As the humai.blog writeup put it, that is “more than the market cap of Etsy. More than Snap.” And the product Cognition sells is Devin, billed as “the world’s first autonomous AI software engineer.” The pitch in a sentence: point it at a Jira ticket, walk away, and the code, the tests, and the pull request show up while a human sleeps. The market is currently pricing in that pitch. Engineers are reading it as an eviction notice. Boards are asking founders if they should freeze the headcount plan. None of those reactions are crazy. They are also, mostly, wrong. Why it sounds right The replacement story sounds true because three things are simultaneously real. Coding agents have gotten dramatically better in the last twelve months. Senior engineers I talk to are doing in an afternoon what they used to scope at two weeks. And the customer logos on the autonomy side are not theoretical. According to the same humai.blog piece, Devin’s customer list reads like a “Goldman Sachs cocktail party”: Goldman Sachs itself, Citi, Dell, Cisco, Palantir, Ramp, Nubank, and Mercado Libre. When Cognition shows up at $25 billion in seven months, having gone from $1 million in ARR to $73 million inside nine months, the smoke-or-fire question feels settled. Something is on fire. The instinct to ask “do I still need engineers” is rational once you have seen those numbers in the same week. The instinct is also incomplete. Because the sa