Google Loses Chief Scientist Jeff Dean as AI Talent Exodus Widens

Google is facing a deepening departure of senior AI researchers even as its cloud business posts some of its strongest revenue growth on record, underscoring a widening tension at the center of the $4 trillion company between commercial returns and frontier research.

Jeff Dean, one of the most recognized names in modern machine learning, announced his departure after 27 years at Google. Dean is leaving alongside prominent colleagues Sanjay Ghemawat, Oriol Vinyals, and Quoc Le to launch a new startup called Discovery Loop. In a post on X, Dean described the venture — backed by Google — as a public benefit corporation "whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress."

The announcement came roughly a day after Google reported 82% revenue growth in its cloud division for the second quarter, a figure that has reinforced the company's case for prioritizing infrastructure, efficient models, and enterprise AI services over pure research bets.

Separately, Demis Hassabis is stepping down as CEO of Google DeepMind to become chairman of the unit while taking on a newly created role as chief scientist at Alphabet, where he will focus on longer-term research and the societal implications of artificial general intelligence. Koray Kavukcuoglu, DeepMind's technology chief and Alphabet's chief AI architect, will assume daily management of the division and lead development of the next Gemini model. According to a person close to the DeepMind team, Kavukcuoglu had already been taking on a broader set of responsibilities from Hassabis over the past year, including directing model development and presenting major Gemini releases.

Hassabis also plans to devote more time to Isomorphic Labs, the AI drug-discovery company that grew out of DeepMind.

Dean's exit is part of a broader pattern. All eight authors of the landmark 2017 paper "Attention Is All You Need" — the technical foundation for modern generative AI — have now left Google. Noam Shazeer, one of those authors, departed for OpenAI in June, less than two years after Google paid nearly $3 billion to bring him back through an acquihire. Nobel laureate John Jumper left DeepMind for Anthropic shortly thereafter.

Some researchers inside Google have grown frustrated over access to computing capacity while watching Google Cloud sell its tensor processing units — the company's homegrown AI chips — to outside customers, including Anthropic, according to people familiar with the matter who asked not to be named due to confidentiality.

Gil Luria, an analyst at D.A. Davidson who recommends holding Alphabet stock, said the talent trend reflects a values gap rather than a compensation problem. "They're not interested in commercializing AI," Luria said. "They're interested in being part of history, and so they look at Anthropic, OpenAI or another startup as being the place where they can pursue history."

Tomasz Tunguz, founder of Theory Ventures, offered a counterpoint that aligns with Google's current strategic direction. "I think we are at that place with AI, particularly for a lot of white-collar work, where many of the models that are reasonable are good enough," Tunguz said. "The next evolution of models are likely to be helpful in domains where you have really fancy computers."

Alphabet CEO Sundar Pichai said on the company's most recent earnings call that 90% of Fortune 100 companies are using Gemini Enterprise, reinforcing the commercial traction Google has built even as researchers depart. Alphabet shares are up 16% this year following a 65% gain in 2025, though the stock fell after the latest earnings report on capital expenditure concerns and declined again following Wednesday's organizational announcement.

How Google manages the divide between its cloud-first commercial ambitions and its ability to retain researchers focused on foundational science will be among the more consequential questions in AI over the months ahead.

Disclaimer