AI

Hassabis moves up and out of the DeepMind CEO chair as Google reshuffles AI leadership

It is a restructuring, not a departure: Hassabis keeps a strategic role and continues to run Isomorphic Labs. But the change lands while Google is under pressure to ship a frontier model, and it costs the company a founder-level scientist.

2 min read

By GTO6 NewsdeskNews Desk
GTO6 editorial illustration: an org chart with the top node moved aside to a chair position
GTO6 illustration
Share
Share
In this article

Google has restructured the leadership of its AI research organisation. Demis Hassabis, who co-founded DeepMind in 2010 and has run it since, is moving to a chairman role at the lab and taking the title of chief scientist for Alphabet, Google's parent company. Operational control of DeepMind passes to Koray Kavukcuoglu, until now the lab's chief technology officer, who will report to Google chief executive Sundar Pichai.

Separately, Jeff Dean — one of Google's longest-serving and most senior engineers, most recently a chief scientist on the AI side — is leaving with a group of colleagues to start a new company that Google is expected to back.

What is changing, and what is not

This is not Hassabis leaving Google. He keeps a strategic seat, continues to lead Isomorphic Labs, the drug-discovery spin-out, and by his own account wants to spend his time on longer-horizon questions about artificial general intelligence rather than day-to-day management. Kavukcuoglu is an internal promotion, not an outside hire, which points to continuity rather than a change of direction.

Why the timing matters

The reshuffle arrives at an awkward moment. Google's rivals have set a fast release cadence, and coding performance in particular has become the yardstick frontier labs are judged by. Google has not shipped a new flagship model since early in the year. Handing operational control to the CTO is a reasonable move if the priority is shipping; losing a scientist of Jeff Dean's standing at the same time is a real cost, whatever the strategic logic.

The read

Large research organisations tend to reorganise when execution, not ideas, is the problem. Nothing here suggests DeepMind is short of research talent or ambition. It does suggest Google has decided the next phase is about getting models out the door, and has arranged its leadership to match.