Demis Hassabis Steps Down as Google DeepMind CEO, Becomes Alphabet's Chief Scientist

The person most associated with Google’s AI ambitions is no longer running the lab he built. Demis Hassabis, the co-founder of DeepMind and a 2024 Nobel laureate in Chemistry for his AlphaFold work, has stepped back from the CEO role at Google DeepMind. He isn’t leaving — he’s moving up and sideways into two new positions that are, in theory, more consequential. But the timing, the context, and the names involved make this more than a tidy succession story.

What exactly changed on August 5?

Hassabis assumed two new titles: Chair of Google DeepMind and Chief Scientist of Alphabet, the parent company. Both are newly created positions. He also continues to lead Isomorphic Labs, Alphabet’s AI drug-discovery subsidiary that applies DeepMind’s protein-structure breakthroughs to pharmaceutical research.

The shift formalizes something that had already been happening in practice — Hassabis spending more time on Isomorphic Labs and on broader AGI strategy rather than the daily machinery of running one of the world’s largest AI research organizations. In a note to staff, Hassabis wrote that he believes AGI is close at hand and that getting the next steps right is critical for humanity. That framing — positioning the move as a deliberate pivot toward the most consequential work, not a retreat from operations — is the version Alphabet wants the market to absorb.

Replacing him in daily operations is Koray Kavukcuoglu, previously Google DeepMind’s Chief Technology Officer. Kavukcuoglu was promoted to Senior Vice President and now runs the lab day-to-day. He’s a deep learning researcher who has been at DeepMind since its pre-Google acquisition days and is well regarded inside the lab, but he is stepping into a role that will be judged against the benchmark Hassabis set — and the results he’s expected to deliver on Gemini and whatever comes after it.

Why does this matter beyond an org chart reshuffle?

Because it didn’t happen in isolation. The same reorganization saw Jeff Dean depart Alphabet entirely. Dean — a Google lifer, the engineer behind MapReduce and TensorFlow, and the person who for years held the title of Chief Scientist overseeing Google’s AI research direction — left to start his own company. He took several high-profile colleagues with him.

Losing Hassabis from the CEO chair and Dean from the company in the same announcement is not a routine refresh. These were the two most visible technical leaders in Google’s AI strategy. One moved up; the other walked out the door. The market’s reaction was immediate: Alphabet shares fell 4% on the news.

What was really going on inside DeepMind?

The official narrative centers on Hassabis wanting to focus on the long game — AGI strategy, Isomorphic Labs, the kind of work a Nobel laureate is uniquely positioned to guide. But a Fortune investigative piece published on August 10, five days after the announcement, painted a more complicated picture.

According to Fortune’s reporting, the leadership change followed a period of low morale, a talent exodus, and model delays inside Google DeepMind. The details of which models were delayed and by how much were not fully specified, but the reporting described an environment where internal frustration had been building for months — the kind of slow-burn organizational stress that doesn’t announce itself with a single incident but accumulates until a structural response becomes unavoidable.

It’s worth being precise about what this is and isn’t. Fortune’s account is investigative reporting, not an Alphabet press release. The company’s own framing of the reshuffle emphasized strategic elevation, not internal dysfunction. Both versions can contain truth simultaneously: Hassabis may genuinely want to focus on AGI and drug discovery, and the lab may genuinely have been struggling under the weight of shipping competitive products on Google’s timeline. Leadership changes at this level rarely have a single cause.

Who is Koray Kavukcuoglu, and what does he inherit?

Kavukcuoglu is a deep learning researcher who joined DeepMind before Google acquired it in 2014. He has been CTO through the lab’s most productive stretch — AlphaFold, Gemini, the integration of DeepMind’s research into Google’s commercial products. He is respected as a technical leader, and his promotion signals that Alphabet wants operational continuity rather than a fresh outside hire.

What he inherits is a lab at a crossroads. Google DeepMind is simultaneously a fundamental research organization — the kind of place that produces Nobel Prize-worthy science — and the engine behind Google’s most commercially important AI products. Those two missions pull in different directions. Research wants time, patience, and tolerance for dead ends. Product wants shipping dates, benchmark wins, and competitive parity with OpenAI’s latest release.

What does Jeff Dean’s departure signal?

Dean’s exit is, in some ways, the more telling data point. Hassabis was elevated — you can read that as a reward, a strategic bet, or a graceful step sideways depending on your priors. Dean left to build something new, and he brought people with him. When a figure of Dean’s stature — someone who has been at Google for over two decades, who helped build the infrastructure that made modern Google possible — decides the next chapter is a startup, it says something about where he thinks the center of gravity in AI is shifting.

The specifics of Dean’s new company haven’t been fully disclosed, but the departure itself is a talent signal that competitors will notice. Senior researchers and engineers deciding where to spend their next five years watch these moves closely. A departure at Dean’s level doesn’t just remove one person from Google’s bench — it opens a door that others may walk through.

What does this mean for Google’s AI competitiveness?

Google is not short on AI talent, infrastructure, or resources. It has the data, the compute, the distribution through Search, Cloud, and Android, and a research pedigree that still runs deeper than any competitor’s. But the last two years have shown that having all the ingredients doesn’t guarantee you ship the best product first. OpenAI’s ChatGPT, Anthropic’s Claude, and Meta’s open-weight Llama models have each forced Google into reactive positions at various points.

The question Kavukcuoglu needs to answer is whether Google DeepMind can operate at both the speed of a product organization and the depth of a research lab — the exact tension that, according to Fortune’s reporting, was already straining the organization under Hassabis. If the answer is yes, the reshuffle is a clean handoff. If the answer is no, this is the moment the difficulty became structural rather than personal.

Where does this leave the AI leadership landscape?

The reshuffle reshapes the top of the AI industry’s org chart. Hassabis joins a small group of technical founders who have stepped away from daily operations to occupy strategic roles — a pattern familiar from other industries but still new in frontier AI, where the founding researchers typically remain hands-on through the most consequential development phases.

For Google, the next six months will reveal whether this was a well-timed transition or a forced one. For the broader industry, it’s a reminder that the organizations building the most powerful AI systems in history are still figuring out how to run themselves — and that the people who started these labs don’t always end up being the ones who ship the products that define them.