Google DeepMind Loses Four Top Researchers in Six Days as Rivals Systematically Raid Its Bench In six days, four of the most consequential researchers in the history of Google DeepMind have announced departures to competitors, a brain drain that strikes at the core of the company’s ability to compete at the frontier. First went Noam Shazeer, co-author of the original Transformer paper and a key Gemini architect, now joining OpenAI. Then came Nobel laureate John Jumper, AlphaFold’s lead researcher, heading to Anthropic. Now Bloomberg reports that Jonas Adler, who ran Google’s AI coding effort, and Alexander Pritzel, a pretraining specialist and AlphaFold contributor, are also joining Anthropic. These are not ordinary hires. Shazeer’s departure means OpenAI now has one of the men who invented the foundational architecture that powers every major AI model in existence. Jumper, Adler, and Pritzel together represent the core leadership of the AlphaFold team transferring to Anthropic. That is not poaching at the margins. The reported trigger is revealing: Google internally reallocated compute away from Shazeer’s projects to a London-based team. In frontier AI research, GPU access is oxygen. Restrict it and people leave for labs where it flows freely. Anthropic and OpenAI are offering both pre-IPO equity at triple-digit revenue growth rates and a credible promise of compute access. Google, for all its infrastructure wealth, apparently could not match that combination for four people in the span of one week. Demis Hassabis told reporters movement between labs is expected. Four departures in six days is not movement. It is a statement. Keywords: Google DeepMind researchers leaving, Noam Shazeer OpenAI, John Jumper Anthropic, AI talent war 2026