This week, Anthropic, the AI lab behind Claude Code, announced its first acquisition: Bun, the blazing-fast JavaScript runtime built in Zig. The deal comes as Claude Code hit $1 billion in run-rate revenue just six months after launch. Great news for Anthropic. Slightly awkward news for their messaging.
Here's the thing that's hard to ignore: for months, we've heard prominent voices from Anthropic and other AI labs confidently predict that software engineers will be obsolete within 3-6 months. Not "transformed" or "augmented," just plain old obsolete. The profession will cease to exist, they tell us, replaced entirely by AI agents that can code better, faster, and without needing health insurance.
And yet.
When it came time to scale their billion-dollar coding agent infrastructure, Anthropic didn't fork Bun from GitHub and let Claude handle it. They didn't spin up a swarm of AI agents to maintain and develop the codebase. They didn't even try.
Instead, they wrote a check and bought the entire company, Jarred Sumner, his engineering team, and all the human expertise that built Bun from scratch.
The Open Source Irony
Bun is MIT-licensed. It's sitting right there on GitHub with over 82,000 stars and 7 million monthly downloads. The runtime is known for its blazing speed-Jarred's own observations showed JavaScriptCore starting around 4x faster than V8, a key advantage in the toolchain wars:
Anthropic could have forked it (perfectly legal) and pointed their supposedly omnipotent AI agents at it.
Had those agents maintained, improved, and scaled it, and saved whatever they paid for the acquisition.
But they didn't.
Mike Krieger, Anthropic's Chief Product Officer, explained it clearly: "Jarred and his team rethought the entire JavaScript toolchain from first principles while remaining focused on real use cases." Translation: We need the humans who understand this deeply.
The announcement emphasized bringing "the Bun team into Anthropic," not the codebase, not the license, but the team. The humans. The software engineers.
What They Say vs. What They Do
There's a cognitive dissonance here that's hard to ignore. When talking about the future, AI labs paint a picture where software engineering as a profession vanishes within months. When talking about their own infrastructure needs-the critical systems that keep their billion-dollar revenue streams flowing-they acquire engineering teams.
Anthropic had been using Bun for months before the acquisition. They knew it worked. They knew the codebase. They had already integrated it into Claude Code's architecture. If AI agents could truly replace software engineers right now, that would be the perfect proof point. Fork it, agent it, ship it.
Instead, they effectively said: "We're growing so fast that we need to own the team that built this."
The Real Message
Look, I don't think Anthropic or anyone else is being deliberately dishonest. But there's a difference between what AI will do eventually and what it can do today. And there's an even bigger gap between what companies claim AI can do and what they trust it to do with their own infrastructure.
When Bun's founder Jarred Sumner wrote about joining Anthropic, he noted they had "over 4 years of runway to figure out monetization." They didn't need to sell. But Anthropic wanted the team badly enough to acquire them anyway, to have those engineers in-house, working directly on the infrastructure that powers Claude Code.
That's not the behavior of a company that believes software engineers are about to be obsolete. That's the behavior of a company that considers expert software engineers so valuable that you should buy entire companies to get them on your team.
The Takeaway for Working Engineers
Should we be worried about AI disrupting software engineering? Absolutely. It already is. Should we believe we'll all be out of jobs in six months? The Bun acquisition suggests we should take those predictions with a massive grain of salt.
Watch what they do, not what they say. And what Anthropic just did was buy a company full of software engineers because they needed human expertise to scale their AI coding infrastructure.
Software engineering isn't as obsolete as the prophets claim. Or maybe it is, and Anthropic just hasn't told their acquisition team yet.
The author is a senior software engineer who remains stubbornly non-obsolete, despite what the timeline says should have happened by now.
