Alphabet has reorganised the leadership of its artificial intelligence operations, moving Google DeepMind CEO Demis Hassabis into a wider strategic role as the technology group attempts to balance frontier research with the growing commercial demands surrounding Gemini.
Hassabis will become chief scientist of Alphabet and chair of Google DeepMind, relinquishing the day-to-day operational responsibilities he has held at the influential AI laboratory. However, he is neither retiring nor leaving Google.
Koray Kavukcuoglu, Google DeepMind’s chief technology officer and Google’s chief AI architect, will take operational control of the organisation as senior vice-president of Google DeepMind. Notably, Google has not appointed him CEO.
The change means Kavukcuoglu will directly oversee Gemini model development, frontier AI research, the Gemini application and Google’s AI developer teams. He will report to Google and Alphabet CEO Sundar Pichai.
“Demis will become the Chair of GDM and Chief Scientist of Alphabet, while continuing to lead Isomorphic Labs,” Google and Alphabet CEO Sundar Pichai said while announcing the leadership changes.
Google DeepMind Leadership Changes at a Glance
| Executive | Previous position | New position and responsibilities |
|---|---|---|
| Demis Hassabis | CEO, Google DeepMind | Chair of Google DeepMind and chief scientist of Alphabet |
| Koray Kavukcuoglu | CTO, Google DeepMind; chief AI architect, Google | SVP of Google DeepMind; responsible for Gemini, frontier models, app and developer teams |
| Jeff Dean | Chief scientist, Google | Leaving to co-found Discovery Loop |
| Sanjay Ghemawat | Google senior fellow | Leaving to co-found Discovery Loop |
| Oriol Vinyals | Senior AI researcher and Gemini leader | Leaving to co-found Discovery Loop |
| Quoc Le | Senior AI researcher | Leaving to co-found Discovery Loop |
Also Read | Google Pay Launches AI-Powered ‘Ask Google Pay’ to Simplify Personal Finance!
Why Demis Hassabis Is Giving Up Operational Control
Hassabis said he had decided to hand over his management duties because he wanted sufficient time to address the larger scientific and societal questions emerging around artificial general intelligence.
AGI generally describes an AI system capable of performing a broad range of intellectual tasks at or beyond human level, rather than being designed for a limited set of functions.
In a message shared with employees, Hassabis said he had worked towards AGI throughout his life and believed it was now “close at hand.” He described the current period as a pivotal point at which decisions about the development and governance of powerful AI systems could have long-term consequences for humanity.
Under the new structure, Hassabis is expected to focus on:
- Alphabet’s long-term AGI research strategy;
- AI safety and the societal consequences of advanced models;
- Scientific applications of artificial intelligence;
- International engagement concerning AGI development;
- Strategic guidance for Google DeepMind’s research teams; and
- The expansion of Isomorphic Labs’ AI drug-discovery programme.
Hassabis will remain closely involved with Kavukcuoglu and Google DeepMind’s research leadership, advising the company on models and frontier research. Alphabet therefore appears to be separating its immediate AI product execution from its longer-term scientific strategy.
Koray Kavukcuoglu Gets Control of Google’s AI Engine
Although the announcement represents a major change on paper, internal reporting suggests that Kavukcuoglu has already exercised substantial influence over Google’s model-development programme.
He has worked at DeepMind for more than 13 years and helped establish its deep-learning team. His research and leadership have been associated with projects including WaveNet, which advanced computer-generated speech, and Deep Q-Networks, which demonstrated how AI could learn to play games from visual information.
Technology journalist Alex Heath reported that some Google DeepMind employees viewed the transfer as closer to a formalisation of the existing working arrangement than an abrupt regime change. Kavukcuoglu had reportedly been the principal leader handling model development for at least a year, while Hassabis increasingly concentrated on AGI, science and external engagement.
That account aligns with Hassabis’ own statement that Gemini’s models had effectively been in Kavukcuoglu’s hands “for a while.”
Also Read | Google's Spam Update Just Changed The Rules, Here Is What Indian Banks Need To Know
The Unreleased Gemini Models Add Pressure to the Transition
The leadership reset comes during an important phase for Google’s AI model roadmap.
Google has yet to release its anticipated Gemini 3.5 Pro frontier model, and the company did not provide a significant update on it during its recent quarterly earnings call. Reuters reported that a flagship version of the model had been expected earlier, raising questions about whether Google’s development timetable had slipped.
Hassabis nevertheless referred to progress on Gemini 4, suggesting that work on a subsequent generation is already underway.
Google also says its lighter Flash models are seeing strong demand, while the Gemma family of open models has surpassed 900 million downloads. The Gemini app, according to Pichai, has reached more than 950 million monthly users.
These numbers demonstrate distribution at an enormous scale. However, the leadership restructuring shows that reach alone is not Google’s only concern. The company must also demonstrate that it can consistently produce frontier models, deploy them quickly and translate its AI research into commercially successful products.
The Bigger Picture
Demis Hassabis stepping down as Google DeepMind CEO does not mark the end of his influence at Alphabet. His authority is shifting away from operational management and towards questions that could define the company’s next decade: AGI, scientific discovery, AI safety and healthcare.
For Google, the immediate test will be execution. Kavukcuoglu must demonstrate that the new structure can accelerate Gemini without weakening DeepMind’s research culture.
For Hassabis, the challenge is even larger: transforming frontier AI from an impressive digital technology into a scientific tool capable of producing breakthroughs in medicine and other fields.
The reshuffle may ultimately be remembered less as a CEO departure and more as the moment Alphabet separated the race to build better AI products from the longer-term pursuit of machines capable of transforming science itself.
