After 27 Years at Google, Jeff Dean Departs with Three Top Scientists to Launch AI Startup; Kavukcuoglu Takes Over Gemini 4 Development

Deep News03:49

Google is facing a significant exodus of top-tier AI talent that is becoming increasingly difficult to ignore.

Chief Scientist Jeff Dean has announced his departure from Google after a 27-year tenure, leaving to co-found an AI startup called Discovery Loop, which aims to automate scientific research workflows. He is joined by three other distinguished Google scientists: Oriol Vinyals, Quoc Le, and Sanjay Ghemawat—all of whom have achieved milestone-level accomplishments in AI and computer science. Alphabet will provide cloud computing resources to the new company and participate in its investment, with the seed round co-led by Radical Ventures and Khosla Ventures.

Demis Hassabis, head of Google DeepMind, will transition to become chairman of the AI lab. Koray Kavukcuoglu, Google's chief AI architect and Hassabis's long-time deputy, will take over most of Hassabis's responsibilities. This personnel shift occurs against a backdrop of sustained competitive pressure in Google's AI landscape. While Google regained a leading position in frontier model competition last autumn with the new version of Gemini, it has since been overtaken by OpenAI and Anthropic, and the release of its latest flagship model, Gemini 3.5 Pro, has been delayed. Following the announcement, Google's stock price fell as much as 5% during intraday trading.

Four Google Veterans Depart Together

Dean joined Google in 1999 as the company's 30th employee and has since driven several key technological transformations. He co-founded Google Brain in 2011, the foundational team for Google's deep learning research, and led the development of the Tensor Processing Unit (TPU)—a chip that now powers the computing for Google's large language models and has been sold to external companies. The three scientists accompanying him have equally impressive backgrounds. Vinyals is a major contributor to AlphaFold, the AI program for protein structure prediction that won the Nobel Prize in 2024. Le co-authored and guided Google's AI system that achieved a gold medal-level performance in the International Mathematical Olympiad. Ghemawat, another early Google employee, co-built Google's distributed systems infrastructure with Dean in the early 2000s and received a joint award from the Association for Computing Machinery (ACM) in 2012. Google CEO Sundar Pichai said Dean and Ghemawat had driven the company's most significant technological transformations during their 27-year tenure, and he wished them well.

AI Department Restructuring: Hassabis Takes Chief Scientist Role, Kavukcuoglu Leads Gemini 4

Jeff Dean's departure is not an isolated case. Google is facing a series of AI talent losses, and compared to competitors like Anthropic and OpenAI, it has experienced delays in launching top-tier large language models. Against this backdrop, Pichai has also announced a major restructuring of the AI department. Demis Hassabis will hand over day-to-day operational duties, transitioning to become chairman of DeepMind and serving as Alphabet's chief scientist. In a post on X, he stated, "I have worked toward artificial general intelligence (AGI) my entire life; as we enter this critical moment, I will focus on long-term strategy and accelerating scientific breakthroughs." He added that he would work with Pichai on "strategic and global" AI-related matters. Hassabis's upward move hands the technical execution baton to Koray Kavukcuoglu. As DeepMind's technical lead, Kavukcuoglu will report directly to Pichai and take on the development of Gemini 4. With the generational gap between the Gemini series and competing products still not closed, the managerial change in the midst of development introduces new uncertainty into the project's progress.

Automated Scientific Experimentation, Targeting Drug Discovery and Clean Energy

Discovery Loop is positioned as a public-benefit corporation, with its core mission to accelerate discoveries by automating complex, multi-step research processes in science and engineering through machine learning models. "The scientific process is coming up with an experimental idea, setting up and running the experiment, and then evaluating the results," Dean said in an interview. "In many fields, you can fully computerize this entire loop." This is the origin of the company's name. Discovery Loop plans to first focus on automating machine learning research and engineering using its own technology stack, and will gradually expand to global challenges such as hardware design, drug discovery, and clean energy. Regarding the reason for leaving Google, Ghemawat noted that Google's existing infrastructure is well-suited for large-scale consumer applications, advertising systems, and search, but is "fundamentally different from the requirements" for the infrastructure they aim to build for research. Dean added that he began seriously considering this entrepreneurial direction about five weeks ago and quickly contacted the other three partners.

Talent Exodus Tests Google's AI Competitiveness

This is not an isolated incident for Google. Reports indicate that Noam Shazeer, a co-author of the transformer architecture paper and co-founder of Character.AI, recently left Google to join OpenAI. John Jumper, a co-recipient of the Nobel Prize for AlphaFold, has moved to Anthropic. The continued departure of high-end AI research talent, combined with slow progress on Google's frontier models, is deepening external doubts about its AI competitiveness. The release timeline for Gemini 3.5 Pro remains unclear, while some new models from Meta have surpassed Gemini on several benchmark tests, further compressing Google's lead.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment