DeepMind loses dominance in AI talent race as rivals lure researchers

Data from UK-based Zeki Data indicates that Google DeepMind’s grip on top AI talent is eroding, with its European, Middle Eastern and African hiring share plunging from 49% to 18.6% over the past few years. The lab lost long-time chief scientist Jeff Dean, senior fellow Sanjay Ghemawat and several researchers who founded the startup Discovery Loop, while co-founder Demis Hassabis shifted to a chairman role and handed day-to-day control to CTO Koray Kavukcuoglu.

Competing labs such as OpenAI, Anthropic, Meta and Microsoft have been poaching staff with lucrative cash-heavy packages and pre-IPO equity, and a tightening of DeepMind’s internal publication policies has further dampened its appeal. The hiring-to-departure ratio fell from about 12-to-1 in early 2023 to roughly 2-to-1 in late 2026, far behind rivals whose ratios range from 3-to-1 to 22-to-1. Departures have been especially pronounced in large-language-model, multimodal and computer-vision teams, while the AlphaFold group lost 13 of its 29 original authors. Despite the setbacks, DeepMind continues to attract more staff than it loses, leveraging Alphabet’s resources and expanding in robotics and embodied AI.