Google DeepMind Restructures Leadership: Hassabis Steps Down and Kavukcuoglu Takes Over
On August 5, 2026, Alphabet executed the biggest leadership restructuring in Google DeepMind's history: Demis Hassabis steps down from running the lab to become its chair and Alphabet's chief scientist, while Koray Kavukcuoglu takes over daily operations. The move reshuffles the top of Google's AI unit amid an intensifying race with Anthropic and OpenAI.
What Exactly Changed on August 5
The announcement was not a simple change of business cards: it reshapes the chain of command of Google's most important AI lab and redefines the role of its key figures.
Hassabis: From CEO to Chair and Alphabet Chief Scientist
Demis Hassabis, cofounder and CEO of the lab since the merger with Google Brain, leaves operational management. From now on he serves as chair of Google DeepMind and chief scientist of Alphabet, the parent company: he keeps authority over the scientific agenda but hands over day-to-day execution. He also remains linked to Isomorphic Labs, the computational biology company he cofounded.
Kavukcuoglu: Google DeepMind SVP Reporting Directly to Pichai
Koray Kavukcuoglu, until now DeepMind's CTO and Alphabet's chief AI architect, takes over as senior vice president of Google DeepMind. His distinctive feature: he reports directly to Sundar Pichai, Alphabet's CEO, with no intermediaries, shortening the decision chain between the lab and the company's top leadership.
Jeff Dean and the Departure of the Business-Unit Chief Scientist
The reorganization also affected Jeff Dean, who left the chief scientist role at the business-unit level, a position that spanned Google and DeepMind. His departure adds to other executive exits in recent months, within a broader reshuffle of the AI division.
Why Google Is Restructuring Now
The timing is no coincidence: the lab faces the most intense competitive pressure in its history, and Google is responding by shortening decision times.
Competitive Pressure: Anthropic and OpenAI on Benchmarks
Anthropic and OpenAI have taken the lead on key reasoning and agent benchmarks. With a more agile command, Google aims to make product and model decisions without internal friction, something the previous structure made slower.
Chinese Open-Source Models Squeezing from Below
From below, Chinese open-source models compete on price and capability, eroding the advantage of Western labs. The restructuring aims to speed up decision-making to respond on both fronts: quality at the top and cost at the base.
Who Koray Kavukcuoglu Is
13 Years Alongside Hassabis: From DeepMind CTO to Alphabet's Chief AI Architect
Kavukcuoglu is not an outsider reshuffling the lab: he has worked alongside Hassabis for more than 13 years and was DeepMind's CTO during the launch and evolution of Gemini (1.0, 1.5, and 2.x). He knows the technical stack and the roadmap first-hand, which lowers the risk of a rocky transition.
What It Means for Gemini and for Developers
The Model Roadmap: What to Expect
For developers building on Gemini, the change is operational, not directional: the model roadmap continues, with Hassabis overseeing the science and Kavukcuoglu executing. The expectation is a faster release cycle and quicker responses to competition, including pricing pressure from open models.
Hassabis's Role as Chief Scientist and AGI
Hassabis has stated that his new focus is the strategic and global affairs of AGI. As chair and chief scientist, he will keep advising the model and research teams, and his authority over the scientific agenda remains despite the title change.
Context: Talent Exodus and Startups
The restructuring coincides with engineers leaving to found their own startups, a symptom of the talent war defining AI in 2026. Google is betting on a structure that retains its technical leaders and responds faster, at a moment when every quarter of advantage counts in the race for next-generation models.
Conclusion
Google DeepMind's new leadership is not abandoning the AI race: it is reorganizing to run it faster. For users and developers, the signal is clear: Gemini will keep evolving, now with a shorter chain of command and a declared obsession with AGI. Keep reading the blog to stay on top of Google's AI and the rest of the industry.
