Google DeepMind loses top talent to rivals, data shows

The new data from Zeki Data paints a clear picture of a shifting AI talent market. DeepMind, long the dominant destination for elite researchers in Europe, is now losing ground to newer labs like OpenAI and Anthropic, which are growing their research headcount at roughly 97% and 152% annually respectively, versus 27% for DeepMind. The lab's decline coincides with an internal shift toward commercializing Gemini and tighter publication rules, which some researchers say displaced the open-ended science that originally attracted them.

The data also shows that DeepMind is not losing evenly across all fields. While it has a net loss in large language models and multimodal systems, and the AlphaFold team has lost 13 of its 29 named authors, the lab is gaining in robotics, embodied AI, and machine learning for science. This suggests DeepMind is strategically rebalancing its talent toward areas where it sees future growth, even as it cedes ground in core AI research.

For a lab that built its reputation on breakthroughs like AlphaGo and AlphaFold, the challenge is not just hiring but retention. As Hurd noted, the exodus of research-heavy people can compound, as they often leave to start their own ventures and bring colleagues with them. With DeepMind still able to attract staff thanks to Google's resources, the question is whether it can hold onto its remaining top researchers amid intense competition.