Google reorganized the leadership of its entire AI operation this week in a single announcement that also cost it one of its most senior engineers. Demis Hassabis, the Nobel Prize-winning co-founder who has run Google DeepMind since its creation, is stepping back from day-to-day control of the lab to become its chairman and Alphabet’s chief scientist. On the very same day, Jeff Dean, Google’s chief scientist and one of its most storied engineers after 27 years at the company, announced he is leaving to co-found a new startup, taking three other senior researchers with him. Alphabet shares fell as much as 5% on the news.
What actually changed
Koray Kavukcuoglu, DeepMind’s chief technology officer and Google’s chief AI architect for the past 13 years, now runs the lab’s daily operations as senior vice president, reporting directly to Alphabet CEO Sundar Pichai rather than holding a standalone CEO title. He now oversees Gemini model development, frontier AI research, and the Gemini app and developer teams, areas that had been Hassabis’s direct responsibility.
Hassabis is not leaving Google. He keeps his job as chief executive of Isomorphic Labs, DeepMind’s AI drug discovery spinoff, and picks up the additional title of chief scientist for Alphabet as a whole, a role explicitly framed around long-term strategy and scientific breakthroughs rather than shipping products. In a memo, Pichai described it as Hassabis stepping up, not stepping down, and Hassabis echoed that framing on X, saying he was stepping into a new role as Chair of Google DeepMind and Chief Scientist of Alphabet, and that he was excited for Kavukcuoglu to step up and lead the lab as SVP.
People close to the situation describe the shift as less abrupt than the announcement made it look. Semafor had previously reported that Hassabis had been shifting day-to-day responsibility for Gemini and Google’s consumer AI strategy to Kavukcuoglu for at least a year, and a DeepMind employee told TIME that Hassabis had already been largely absent from routine Gemini meetings, attending only the most significant ones. Kavukcuoglu, notably, has also navigated Google’s internal politics more successfully than some peers, according to a source who spoke to Axios.
The departure that overshadowed it
Jeff Dean’s exit is, by most measures, the harder loss. Dean was Google’s employee number 30, joined in 1999, and became one of the most recognizable names in machine learning research over nearly three decades at the company, work that touched everything from Google’s core search infrastructure to TensorFlow to the underlying research that shaped Gemini. He is leaving alongside Sanjay Ghemawat, a longtime collaborator and fellow Google Fellow, along with DeepMind VP Oriol Vinyals and Google Brain co-founder Quoc Le, to launch a new company called Discovery Loop, aimed at automating machine learning, science, and engineering research to accelerate scientific discovery. Google is reportedly a founding investor and cloud partner in the new venture and says it plans to collaborate with Discovery Loop on research infrastructure going forward, an unusually cooperative arrangement for a competitor spun out by departing executives.
A Google spokesperson said the two announcements, Hassabis’s move and Dean’s exit, were not connected, despite landing on the same day. Not everyone outside the company is convinced that timing is coincidental, and even if it is, the optics compound each other regardless of the actual causal relationship. Former Googler Peter J. Liu wrote, in a post later deleted, that the magnitude of this is hard to understate, since Gemini leadership had left, Demis was no longer head of DeepMind, and more talent would likely be leaving. Dan Shipper, CEO of the AI subscription service Every, framed it more bluntly on X, arguing that to be competitive today Google needs to catch up on frontier coding.
A pattern of departures, not an isolated one
Dean and his co-founders are not the first prominent names to leave DeepMind this year. John Jumper, who shared a Nobel Prize with Hassabis for AI-driven protein structure prediction, left DeepMind earlier in 2026 to join Anthropic, among other high-profile departures reported over recent months. That accumulation matters for how this week’s news is being read inside and outside the company: DeepMind has weathered individual exits before, but Hassabis and Dean both carry what TIME described as near-legendary status within Google, and losing both from active operational roles in the same 24-hour window is a different order of event than routine attrition.
Some DeepMind employees reportedly see the restructuring itself, separate from who is leaving, as a further erosion of the unit’s once-independent standing inside Google. Hassabis held the title of CEO of Google DeepMind; Kavukcuoglu is being given the title of senior vice president, one level down in formal organizational terms, reporting into Pichai rather than running an autonomous division. DeepMind was folded into Google’s broader structure once already, when Google Brain and DeepMind merged in 2023 to end a long-running, well-documented rivalry between the two research groups. This week’s changes read, to some observers, as another step in the same direction: less autonomy for the lab that produced AlphaGo, AlphaFold, and the Gemini model family, not more.
Why Hassabis’s own writing matters here
The timing of Hassabis’s shift toward a strategy-and-science mandate lines up with an essay he published in July, a Framework for Frontier AI, in which he argued AGI is probably only a few short years away and called for periodic independent evaluations and sector-specific regulation. That’s a governance-facing, industry-level agenda, closer to what a chief scientist or elder statesman of the field would pursue than what a product-shipping CEO spends their days on. Moving into an Alphabet-wide advisory and science role, rather than leaving the company outright, lets Hassabis keep making that broader case publicly while Kavukcuoglu runs the actual Gemini roadmap day to day. Read that way, the reorganization is not purely defensive, it also formalizes a division of labor that had reportedly already been happening informally for a year.
Both Hassabis and Pichai released upbeat memos framing the changes as strength rather than retreat. Pichai wrote that Google has amazing talent, world-class compute, and products that bring AI to more people than any other company, while also acknowledging the need to accelerate that work and stay focused on the AI frontier, language that implicitly concedes Google feels pressure to move faster. Hassabis, for his part, tied his new role directly back to his longest-standing stated priority, saying he has always believed the number one application of AI should be to improve human health, and that it is time for AI to prove its value to the world by helping finally cure diseases like cancer.
What this means going forward
For anyone tracking Gemini’s product direction, the practical answer is that Kavukcuoglu was already the person most directly shaping recent releases, so near-term continuity is likely. The bigger open question is talent retention. Google has absorbed individual high-profile exits before without major disruption, but a lab that just lost its CEO from day-to-day operations and its most senior technical fellow in the same week, on top of a year of accumulating departures to competitors like Anthropic, is now the subject of real, public scrutiny over whether more senior researchers follow Dean out the door. The market’s initial reaction, a same-day stock drop of up to 5 percent, suggests investors are treating that risk as real rather than theoretical, at least until Google’s next several product releases and retention numbers say otherwise.
The precedent this echoes
Google has restructured its AI organization before, and the last time it did so was itself the product of a competitive scare. Google Brain and DeepMind operated as separate, often rival research groups for years before merging in 2023, a consolidation widely read at the time as a response to the sudden threat OpenAI’s ChatGPT posed to Google’s core search business. That merger was framed then, much as this week’s changes are being framed now, as a move toward speed and unified execution rather than a sign of internal weakness. Whether this week’s changes ultimately read the same way in hindsight will depend heavily on what Google actually ships over the next several quarters, and on whether Kavukcuoglu’s more hands-on operational style produces a faster cadence of releases than the structure it replaces.
What Discovery Loop means for the competitive field
Dean’s new venture is worth watching independently of the Google story it’s tangled up with. Discovery Loop is aimed specifically at automating machine learning, science, and engineering research itself, a meta-level bet on using AI to accelerate the process of discovering new AI techniques and scientific results, rather than building consumer-facing products directly. That’s a genuinely different lane from most well-funded AI startups launching this year, and the fact that Google is reportedly staying on as both an investor and cloud partner despite Dean’s departure suggests Google sees more value in maintaining a research relationship with him than in treating his exit as purely adversarial. For a company with Dean, Ghemawat, Vinyals, and Le’s combined track record, expectations for what Discovery Loop actually produces will be high from the moment it launches publicly.
See TIME’s original reporting for additional detail on the internal dynamics behind the shift.
