Two senior Gemini researchers set to leave Google for Anthropic as AI talent drain continues

Summary

Google Deepmind is set to lose Jonas Adler, who led AI coding at Gemini, and Alexander Pritzel, a training researcher, to Anthropic, Bloomberg reported on June 24. The departures add to a succession of high-profile exits: John Jumper, the DeepMind director and AlphaFold Nobel co-recipient, has already joined Anthropic, and Noam Shazeer, a Google employee of more than 20 years and co-founder of Character.AI, is reportedly heading to Openai. Bloomberg and TechCrunch both cited pre-IPO equity at Anthropic as the primary pull factor, with Anthropic's Series H funding round in late 2025 having set the stage for a potential listing.

Why it matters

The pattern is structural, not incidental: Anthropic was built by former Google Brain researchers and has since pulled back researchers from the same pipeline at an accelerating rate as its IPO approach makes pre-IPO equity attractive. Losing both a Gemini coding lead and a training researcher within the same week signals the drain is reaching the technical core, not just adjacent roles. The timing pressures Google Deepmind precisely when Gemini 2.x differentiation against Claude and GPT-4o is still unsettled.

What to watch

• Whether Google responds with accelerated vesting or new retention mechanisms tied to Gemini 2.x milestones.
• Anthropic's IPO timeline: if it lists before OpenAI, it crystallises the equity thesis that is driving departures.
• Whether Adler and Pritzel roles at Anthropic signal where the company is investing, specifically in coding models or training infrastructure.