
Google Deepmind loses both its CEO and chief scientist as Demis Hassabis and Jeff Dean step down simultaneously
Quick Answer
Google DeepMind undergoes significant leadership changes as CEO Demis Hassabis transitions to Alphabet's Chief Scientist and Jeff Dean departs to launch Discovery Loop, a new venture aimed at automating scientific research.
Quick Take
This shift comes as Google faces competitive pressure in AI, particularly with its Gemini model development.
Key Points
- Demis Hassabis steps down for a strategic role, focusing on AGI and health applications.
- Jeff Dean leaves after 27 years to co-found Discovery Loop, targeting automated scientific research.
- Koray Kavukcuoglu takes over daily operations at Google DeepMind, overseeing Gemini and AI research.
- Discovery Loop aims to automate the research cycle, addressing major scientific challenges.
- Google's Gemini 4 model is in training to improve competitiveness against OpenAI and Anthropic.
DeepSignal Analysis
What happened
Google DeepMind is experiencing significant leadership changes as CEO Demis Hassabis transitions to Alphabet's Chief Scientist role, while Jeff Dean departs after 27 years to co-found Discovery Loop, a venture focused on automating scientific research. Koray Kavukcuoglu will take over day-to-day management of DeepMind, overseeing key AI projects.
Key evidence
- Demis Hassabis is stepping away from daily operations at Google DeepMind to focus on broader AGI matters as Alphabet's Chief Scientist.
- Jeff Dean is leaving Google to co-found Discovery Loop, which aims to automate scientific research and machine learning experiments.
- Koray Kavukcuoglu, previously DeepMind's chief technology officer, will manage the company and report directly to Google CEO Sundar Pichai.
Why it matters
These leadership changes come at a critical time for Google as it faces increasing competition in the AI sector, particularly with its Gemini model development. The departure of influential figures like Dean could impact Google's innovation trajectory and its ability to compete with rivals like OpenAI and Anthropic. The focus on automating scientific research may also shift the landscape of AI applications in various fields.
📖 Reader Mode
~3 min readGoogle Deepmind is going through a major leadership overhaul. Demis Hassabis is stepping away from day-to-day operations to become Alphabet's Chief Scientist, while chief scientist Jeff Dean is leaving the company after 27 years to start his own AI venture.
Hassabis said the move will give him more time to focus on "the big picture." In a message to employees, he described this as "a pivotal moment in human history." Hassabis said he has spent his life working toward AGI and now believes it is "close at hand." He previously described humanity as "standing in the foothills of the singularity."
Hassabis will continue to work closely with Google CEO Sundar Pichai on "strategic and global AGI matters" while advising Deepmind's teams. He will remain based at the company's London headquarters and also devote more time to Isomorphic Labs, Alphabet's AI drug discovery subsidiary. Hassabis has said that "the No. 1 application of AI should be to improve human health."
Koray Kavukcuoglu will take over day-to-day management of Google Deepmind. He previously served as Deepmind's chief technology officer and Google's chief AI architect. As senior vice president of Google Deepmind, Kavukcuoglu will report directly to Pichai and oversee Gemini model development, frontier AI research, the Gemini app, and Google's AI developer platforms.
Kavukcuoglu joined Deepmind in its early days and has worked with Hassabis for more than 13 years. Pichai said Kavukcuoglu built Deepmind's deep learning team and helped drive breakthroughs such as WaveNet and DQN. Hassabis said he has "total confidence" in Kavukcuoglu and the rest of the leadership team.
Jeff Dean is leaving Google to automate scientific research
Google DeepMind is also losing Jeff Dean, one of the most influential engineers in the company's history. After 27 years at Google, Dean is joining Google Senior Fellow Sanjay Ghemawat to launch Discovery Loop, a public benefit corporation that plans to automate machine learning, science, and engineering to speed up scientific discovery.
Discovery Loop will initially focus on automating large-scale machine learning experiments. The founders eventually plan to apply the same approach across a wide range of scientific and engineering fields. Anthropic and OpenAI have also made faster, more automated scientific research part of their broader AI plans.

Dean and Ghemawat helped build Google's early search infrastructure and later worked on the neural networks that, in Pichai's words, "helped create the modern AI era." Discovery Loop's other co-founders are Oriol Vinyals and Quoc Le, two other longtime Google AI researchers. Dean said the four founders have worked together for periods ranging from 14 to 30 years.
Radical Ventures and Khosla Ventures are leading the seed round, with backing from Lightspeed, Kleiner Perkins, Doerr Capital, and Alphabet. The round is expected to close in the coming weeks. Alphabet will remain involved as Discovery Loop's cloud partner and plans to work with the startup on a research framework for machine learning systems and infrastructure.

Dean's announcement on X doesn't thank Google or mention his 27 years at the company. Prominent tech departures usually come with carefully prepared farewell statements, making the omission unusual. It suggests there may be more to Dean's exit as Google faces mounting pressure in AI, particularly in coding, where its best models have fallen behind those from OpenAI and Anthropic. Google is counting on Gemini 4, a "much larger" foundation model now in training, to close that gap. The company hasn't announced a release date.
— Originally published at the-decoder.com
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.

