Google has confirmed a leadership shake-up at DeepMind, with founder Demis Hassabis moving from chief executive to chair and chief scientist at parent company Alphabet. The change, announced internally by CEO Sundar Pichai and disclosed on Wednesday, hands day-to-day control of the AI lab to longtime CTO Koray Kavukcuoglu, who will serve as senior vice president. The restructuring is the clearest signal yet that Google is recalibrating its AI organization after a stretch of product delays and high-profile departures.
In a note to staff, Hassabis framed the shift as a way to “focus on the big picture and help influence what is to come to the best of my ability.” He has spent recent months working on safety and AGI governance, and said publicly that artificial general intelligence is “close at hand,” making strategic direction more urgent than operational oversight.
Context: A Lab Under Pressure
The reorganization arrives against a backdrop of competitive strain. Gemini, Google’s flagship model family, has suffered repeated delays and is widely viewed as having fallen behind frontier offerings from OpenAI and Anthropic. Talent losses have compounded the problem. Earlier this year, Nobel co-laureate John Jumper, recognized for protein-structure prediction work, left for Anthropic. Longtime chief scientist Jeff Dean and several colleagues have also departed to launch a new venture. The bench that DeepMind built over a decade is being picked apart by rivals flush with capital.
Google’s market reaction was swift. Shares traded lower on the news, reflecting investor concern that the change introduces uncertainty at a moment when the company can least afford it.
Analysis: What the New Structure Changes
Kavukcuoglu is not a newcomer to the top job in practice. According to reporting from Semafor, he has been running Gemini day-to-day for at least a year, which suggests Wednesday’s announcement formalizes an arrangement that was already in motion. Elevating him to SVP without the CEO title is a notable choice: it preserves Hassabis‘ strategic authority through the chair role while signaling that operational execution is the priority. It also leaves Alphabet with more direct leverage over the lab’s roadmap.
The supporting cast is also being rearranged. SVP James Manyika is expected to absorb parts of the reporting line previously held by chief operating officer Lila Ibrahim, and some safety functions could shift toward legal and policy chief Kent Walker. Co-founder Shane Legg, who serves as chief AGI scientist, will continue to report to Hassabis, preserving the chain of accountability for the lab’s most ambitious technical bets.
The personnel moves suggest Google is trying to separate two functions that had been compressed into the CEO role: shipping competitive products on a frontier cadence, and stewarding long-horizon research and safety work. By splitting those responsibilities, the company is betting it can accelerate the first without diluting the second.
Implications: Safety, Competition, and AGI Timing
Google was quick to defend its safety posture. A spokesperson said the transition “changes absolutely nothing” on frontier AI safety, and Kavukcuoglu’s representatives emphasized that “advancing the frontier of AI and building it responsibly are the exact same mission.” Whether that rhetoric holds will depend on how safety teams are ultimately placed in the new reporting structure and whether Hassabis retains enough organizational weight to influence research direction from the chair seat.
For competitors, the reshuffle is an opening. OpenAI and Anthropic have been winning talent and mindshare, and any execution stutter at DeepMind during the transition could widen that gap. For Google, the bet is that clearer ownership of Gemini, combined with Hassabis‘ continued involvement at the strategic level, will close it. The company’s public framing positions this as evolution rather than disruption, but investors clearly read it as a moment of vulnerability.
Why the founder-step-aside framing matters now
DeepMind was founded in 2010 in London and acquired by Google in 2014, and Kavukcuoglu joined via the 2016 Vision Factory acquisition. Gemini launched in December 2023 as a response to OpenAI’s GPT-4 and has since gone through multiple major version updates, while Anthropic, founded in 2021 by former OpenAI VP of research Dario Amodei, has built Claude into a frontier competitor. That history frames why a founder stepping aside at this moment carries weight: the lab is no longer a scrappy startup but a core Alphabet asset whose setbacks are now measured against well-funded rivals.
Hassabis co-founded DeepMind in 2010 in London, and co-founder Shane Legg still leads its chief AGI scientist role, meaning both remaining founders now sit above the operational layer rather than inside it. His “AGI close at hand” framing is unusual for a sitting CEO; rivals such as Anthropic and OpenAI have confined similar “near-term AGI” language to research papers rather than executive messaging. The shift lands as Google faces antitrust scrutiny over search advertising and EU regulatory pressure, making the founder’s elevated but removed posture a deliberate signal about where strategic authority now resides.
Longer term, the move is a bet that AGI is near enough to demand dedicated executive attention, and that operational AI work is mature enough to run without the founder at the helm. Whether Hassabis‘ shift from operator to strategist sharpens Google’s frontier push or accelerates its talent drain will be the story to watch over the next two product cycles. Source: TIME