Google AI chief Jeff Dean departs after 27 years amid talent exodus

Jeff Dean announced his departure as Google's chief scientist on August 6, 2026, after 27 years at the company, part of a leadership shakeup in its artificial intelligence division. Dean is leaving with Sanjay Ghemawat, Oriol Vinyals, and Quoc Le to found Discovery Loop, a Google-backed public benefit corporation with the mission of automating machine learning and scientific discovery.

The departures follow the exit of prominent researchers, including Noam Shazeer, who left for OpenAI in June 2026, and Nobel laureate John Jumper, who joined Anthropic. All eight authors of the landmark 2017 paper 'Attention Is All You Need' have now left Google.

Demis Hassabis stepped down as CEO of Google DeepMind to become the division's chairman and assume the new role of chief scientist at Alphabet, focusing on longer-term research and the implications of artificial general intelligence. Koray Kavukcuoglu, DeepMind's technology chief, will take over daily management and lead development of the next Gemini model.

Google reported 82% revenue growth in its cloud division, underscoring tensions between commercial success and research ambitions. On the July 2026 earnings call, Alphabet CEO Sundar Pichai said 90% of Fortune 100 companies use Gemini Enterprise, highlighting strong enterprise demand.

Tomasz Tunguz, founder of Theory Ventures, noted that top-tier models may be unnecessary for most enterprise needs, saying, 'many of the models that are reasonable are good enough.'

Some researchers have grown frustrated over limited access to computing capacity, especially as Google Cloud sells TPUs to outside customers like Anthropic, according to people familiar with the matter. Tensions rise when Google announces large infrastructure commitments to competing labs.

On the July 2026 call, Pichai reiterated that DeepMind's compute needs remain a 'first priority' to compete at the frontier in AGI development, even as cloud demand grows.

Gil Luria, an analyst at D.A. Davidson, said departing researchers 'are not interested in commercializing AI' but rather in being part of history at startups. Dan Niles, founder of Niles Investment Management, noted that the most powerful models are unnecessary for many commercial applications, but for scientists seeking the next breakthrough, being good enough is not always enough.