Google DeepMind's reorg and why AI might be a bubble

Google is reshuffling DeepMind under a new head, a move that signals how the company's trying to keep the lab from running off on its own. The reorg comes as the broader AI sector faces mounting pressure: big capital has poured in, valuations have stretched, and now the market is asking whether the growth story holds up.

The contrast is sharp. DeepMind is a world-class research outfit, but the companies building products on top of its models are racing to ship. That tension — pure science versus product velocity — is one of the defining friction points in the industry right now. Meanwhile, the dot-com comparison is doing a lot of work in the air. Yes, speculation ran wild in the late 1990s. But the difference is that this time the underlying tech is actually useful: LLMs are a real tool people use, not just a stock ticker.

Why this matters for us: the bubble talk matters because when capital dries up, the companies that can actually make things for working people — the ones that aren't just chasing valuation — are the ones that survive.