Skip to content
IFAI

Industry

Google DeepMind shakes up its leadership as Hassabis steps back

Published Reach 54

Google just announced a major leadership shakeup across its AI ranks, with Demis Hassabis stepping back from day-to-day operations to become Chair of DeepMind and Alphabet’s Chief Scientist. DeepMind veteran Koray Kavukcuoglu takes over as SVP to steer Gemini model development, while legendary 27-year Google veteran Jeff Dean is departing to launch an independent AI-for-science venture. With Gemini topping 950M monthly active users, Google is reorganizing its top minds for the next sprint toward AGI.

Google DeepMind shakes up its leadership as Hassabis steps back
Image: blog.google

Google is shaking up its AI command center just as its flagship products hit unprecedented scale.

Alphabet CEO Sundar Pichai and Google DeepMind co-founder Demis Hassabis announced a sweeping leadership overhaul. Hassabis is stepping back from day-to-day operations to become Chair of Google DeepMind and Chief Scientist of Alphabet. Meanwhile, 27-year Google veteran Jeff Dean is departing alongside legendary engineer Sanjay Ghemawat to launch an independent venture.

Shifting gears from research lab to shipping machine

Taking over the operational helm is Koray Kavukcuoglu, a 13-year DeepMind veteran who previously served as CTO and Chief AI Architect. Kavukcuoglu, who helped build foundational breakthroughs like WaveNet and DQN, steps up to SVP of Google DeepMind, reporting directly to Pichai.

His mandate is massive: oversee Gemini model development, frontier AI research, and the teams building the Gemini consumer app and developer ecosystem.

For Hassabis, the move frees him to focus on the big picture. Writing that humanity is standing in the foothills of the singularity, Hassabis said he will advise on high-level AGI strategy and frontier models like the upcoming Gemini 4 from the new London Platform 37 hub. He will also double down on Isomorphic Labs, reiterating his conviction that AI's greatest test is curing diseases like cancer.

Jeff Dean exits as Gemini nears 1 billion users

The shakeup also marks the end of an era. Jeff Dean, Google's 30th employee and one of the primary architects of its technical infrastructure, is leaving with longtime collaborator Sanjay Ghemawat to start Discovery Loop. The new public benefit corporation aims to build AI systems that automate problem-solving across machine learning, science, and engineering, with Google backing it as a founding investor and cloud partner.

All of this is happening against a backdrop of eye-popping scale. Pichai revealed that the Gemini app has crossed 950 million monthly active users, while the open-weight Gemma model family has topped 900 million downloads. High-efficiency models like Flash are seeing surging demand alongside the release of its Cyber model.

Why it matters

Google is finally separating its existential long-term bets from its brutal daily product wars.

For the past two years, Google asked Hassabis to do the impossible: serve as the intellectual beacon for AGI while simultaneously slugging it out in the trenches against OpenAI to ship consumer software. Passing operational control to Kavukcuoglu solves this bottleneck. With the Gemini app knocking on the door of 1 billion monthly users, Google needs an elite systems builder who can align frontier research with hard-nosed consumer execution.

For developers and enterprise customers, this transition should bring cleaner roadmaps and faster turnaround times. Kavukcuoglu knows the deep learning stack inside out, meaning Gemini 4 and supporting developer tools are likely to be tuned heavily for production reliability rather than just academic benchmark glory.

Still, it is hard to ignore the symbolic weight of Jeff Dean stepping out of the building. Even with Google serving as a founding investor in Discovery Loop, losing the engineers who built Google's foundational compute architecture is a real brain drain. Google DeepMind is visibly transitioning from an idealistic research powerhouse into an enterprise product engine, and how it balances that commercial grind with genuine frontier breakthroughs will define its next chapter.

Sources (2)

  1. [1]
    blog.google(opens in a new tab)Originalblog.google
  2. [2]
    theverge.com(opens in a new tab)Reporttheverge.com

Best of AI, daily