Google has announced a significant restructuring of its top artificial intelligence leadership, reshaping the chain of command at Google DeepMind and triggering a major departure from the company. The changes, laid out by Alphabet CEO Sundar Pichai on Wednesday, August 5, 2026, come at a moment when competition in generative AI is intensifying and when Google is racing to integrate AI across its search, cloud, and consumer products. The announcement is being described inside the company as a generational handoff, with some of the most prominent figures in modern AI research moving into new roles or leaving to pursue fresh ventures.
Demis Hassabis gets a new strategic role
Demis Hassabis, the CEO and co-founder of Google DeepMind, is stepping back from the operational leadership of the unit. He will become the chair of Google DeepMind and take on the title of chief scientist at Alphabet. In those roles, Hassabis will remain a central voice in shaping Google’s long-term AI strategy, but he will no longer manage the day-to-day operations of DeepMind’s research teams.
Hassabis will also continue to lead Isomorphic Labs, an Alphabet company focused on using AI for drug discovery. That keeps him close to one of the areas he has long described as the most meaningful application of artificial intelligence. In a message to staff, Hassabis said that AI should be directed at improving human health and accelerating the search for cures to major diseases. “I’ve always believed the No.1 application of AI should be to improve human health,” he said. “It’s time for AI to prove its unequivocal value to the world, and what better way to demonstrate that than to help finally cure diseases like cancer.”
Hassabis added that he will continue to work closely with Alphabet CEO Sundar Pichai on strategic and global AGI matters and will advise Koray Kavukcuoglu and other DeepMind leaders as they take on expanded responsibilities.
Koray Kavukcuoglu takes over DeepMind
Koray Kavukcuoglu, who previously served as DeepMind’s chief technology officer, is being promoted to SVP of DeepMind. In his new position, Kavukcuoglu will report directly to Sundar Pichai and will lead the research organization that was created in 2023 when Google combined its Brain team with DeepMind. He will also retain the title of chief AI architect at Google, a role that puts him at the center of the company’s machine learning infrastructure and model development.
Kavukcuoglu’s elevation signals continuity for the technical direction of DeepMind. He has worked with Hassabis for years and has deep experience in reinforcement learning, neural networks, and large-scale AI systems. His move from CTO to the top operational seat suggests that Google wants to keep its research agenda close to the product group, while relying on Hassabis for high-level vision and external representation.
Jeff Dean and Sanjay Ghemawat leave to start Discovery Loop
The other major development is the departure of Jeff Dean, one of the most important engineers in Google’s history. Dean, who was Google’s 30th employee and had served as the chief scientist of Google DeepMind, is leaving the company along with Google Fellow Sanjay Ghemawat. Together, they are founding Discovery Loop, an AI-focused public benefit corporation. Google will be a founding investor in the new company, according to Sundar Pichai.
Discovery Loop’s stated mission is to build AI systems that can automatically solve hard problems in machine learning, science, and engineering. The company says it wants to accelerate the pace of scientific discovery and deliver technological solutions that improve lives at a global scale. In a statement published on its website, Discovery Loop said: “Our mission is straightforward: we are building AI solutions that can automatically solve important problems in machine learning, science, and engineering. By advancing the pace at which we conduct engineering and scientific discovery, we can bring the benefits of science and technology to the world much faster. Ultimately, our goal is to build AI systems that act as a deeply positive, empowering force for humanity, delivering technology solutions that improve people’s lives on a global scale.”
A shift with deep roots in DeepMind’s history
Hassabis founded DeepMind in 2010 with Mustafa Suleyman and Shane Legg. Google acquired the London-based startup in 2014. The lab gained worldwide attention in 2016 when its AlphaGo system defeated the world champion Go player Lee Sedol, a breakthrough many viewed as a turning point for modern AI. DeepMind later produced AlphaFold, a system that predicted the three-dimensional structure of proteins from amino acid sequences. AlphaFold became one of the most cited scientific achievements in AI, and Hassabis shared the 2024 Nobel Prize in Chemistry for that work.
Over the years, DeepMind has focused on long-term AGI research, reinforcement learning, robotics, health care, and protein folding. In 2023, Google merged its Brain team and DeepMind into a single unit called Google DeepMind, with Hassabis as CEO. That reorganization was intended to consolidate Google’s AI efforts and improve collaboration between research teams and product divisions. The new leadership changes are the next step in that evolution.
Kavukcuoglu, who joined DeepMind years ago, is known for his work on neural network architectures and reinforcement learning. As CTO, he helped oversee the technical infrastructure that powers many of DeepMind’s research breakthroughs. His new role as SVP of DeepMind will test whether he can balance the lab’s long-term research ambitions with the commercial pressures that come with being part of Google.
A new chapter for Jeff Dean
Before joining DeepMind’s leadership, Jeff Dean was a key architect of Google’s early infrastructure. He helped build the search index, the ad-serving system, and the distributed computing frameworks that made Google’s services scale. With Sanjay Ghemawat, Dean created MapReduce, a programming model that became foundational to big data processing. He also helped design BigTable, a distributed storage system, and later became a driving force behind TensorFlow, Google’s open-source machine learning framework.
Dean’s move to Discovery Loop marks the end of an era at Google, where he spent more than two decades as a researcher and technical leader. In recent years, he served as chief scientist at Google DeepMind while also remaining deeply involved in the company’s AI infrastructure. His departure, together with Ghemawat, is a notable loss for Google, but the company’s decision to invest in Discovery Loop keeps a connection to the new venture.
What this means for Google’s AI strategy
The leadership shakeup comes as Google faces intense competition in the AI market. OpenAI, Microsoft, Meta, and many startups are pouring resources into large language models, conversational AI, and agentic systems. Google has responded by integrating Gemini into Search, Workspace, and Cloud, and by building TPUs and other specialized hardware for AI workloads. The new structure puts a larger burden on Kavukcuoglu to execute on these products, while giving Hassabis the freedom to focus on broad technical strategy and the long-term pursuit of AGI.
Hassabis’s continued leadership of Isomorphic Labs is also significant. Drug discovery is a field where AI can deliver tangible medical breakthroughs, and the timing of the leadership change suggests that Google is prepared to invest in applications that go beyond search and generative AI. The company has already used AlphaFold in collaborations with research labs and pharmaceutical companies, and Hassabis’s new title could allow him to push more of that work forward.
Discovery Loop, meanwhile, could become an important new player in the AI research ecosystem. With Jeff Dean and Sanjay Ghemawat leading the team, the company has instant credibility. Its public benefit status signals a commitment to using AI for scientific progress rather than purely commercial gain. Google’s relationship with the company as a founding investor could also create a new channel for research collaboration, especially if Discovery Loop develops tools that are complementary to Google’s own AI stack.
Source:The Verge News