The best AI researchers just switched teams — and it matters to you
When four top DeepMind researchers leave in one week and three join the same rival, that is a clue about which AI tools your team will use next year.
- 4
- top researchers left in six days
- 3
- of them went to Anthropic
- ~6%
- drop in Alphabet's stock in five days
- 200M+
- protein shapes mapped by Jumper's AlphaFold
DeepMind is Google's AI lab. In one week, four of its most senior researchers quit. Reporters covered it like a sports transfer: who left, who poached whom. But there is a bigger point. If you build software on AI, your tools depend on a few people — and a group of them just changed jobs in six days.
The names matter. Noam Shazeer helped invent the "transformer," the technology behind today's chatbots. He left for OpenAI on June 18. The next day, John Jumper left Google after almost nine years and joined Anthropic. Jumper won a Nobel Prize in 2024; his AlphaFold program mapped the shapes of more than 200 million proteins. By June 24, two more researchers had followed him, Jonas Adler and Alexander Pritzel. Three of the four went to Anthropic.
Why did they leave? Partly money. Anthropic and OpenAI both plan to sell shares to the public, so a stake could be worth a lot later — more than a steady salary. There was tension too: reports say computing power was taken from Shazeer's work and given to another team. Money and computers can be bought, but the skill to use them well cannot. Investors noticed — Alphabet's stock fell about 6% over five trading days.
If you build with AI, here is the lesson: do not tie yourself to one AI company. Keep your prompts and setup easy to move, so you can switch to a better model from another lab without rebuilding your product. And watch where top researchers land, not just which model scored best this quarter — the people are the early clue. In AI, you are renting skill from a market that can reshuffle in a week.