Why are Google's top AI researchers leaving? A plain-English guide

If you've seen headlines about Google 'losing' its AI stars, here's what actually happened — and why it's a bigger deal than a normal job change. In just one week in June 2026, four of Google's most respected AI researchers announced they were leaving for two rival labs. When people this senior move all at once, it tells you something about where the industry thinks the future is being built.

Think of it as four exits in seven days. Noam Shazeer — one of the people who invented the 'Transformer,' the core technology behind modern chatbots — is joining OpenAI. John Jumper, who shared a 2024 Nobel Prize in chemistry for AlphaFold (an AI that predicts protein shapes), is joining Anthropic. Two more researchers, Jonas Adler and Alexander Pritzel, are heading to Anthropic as well, and a fifth, Arthur Conmy, said he'll join Anthropic to work on AI safety.

The biggest reason is money that hasn't been paid out yet. Anthropic recently raised $65 billion and is valued at about $965 billion, which makes it the most valuable private AI company in the world — even ahead of OpenAI. Both labs may go public soon. If you join before that happens and receive equity (a share of the company), it could become worth a great deal. That's a powerful draw that a public company like Google finds hard to match.

There were other reasons, too. Reporting says the computing power for one of Shazeer's projects was moved to a different team in London before he left, which points to some internal friction. And Anthropic is expanding into coding, health and science — areas that match what these researchers do best. So it's a mix of pay, workplace tensions and a good fit for their interests.

Not necessarily. Google's AI chief, Demis Hassabis, said at an event in Cannes that talent moves between all the top labs, that Google wins its fair share, and that it still has by far the biggest and broadest research team of any lab. The stock market was a bit more nervous — Alphabet shares dipped about 5-6% on 22 June — but a share dip and a real decline in ability are not the same thing.

It's also worth zooming out on what this means for the AI you actually use. When top researchers cluster at one or two labs, the models those labs ship tend to improve faster, because breakthroughs come from small numbers of exceptional people, not headcount alone. So a wave of talent moving to Anthropic and OpenAI could show up, months from now, as noticeably better coding help, science tools or assistants from those companies. It doesn't mean Google's products will get worse — Google still has a deep bench and enormous resources — but it's a hint about where the next round of impressive releases might come from. If you're deciding which AI tools to invest your time learning, watching where the best people go is a surprisingly good signal, and right now that signal points at the two labs racing towards the public markets.