Google DeepMind Splits Research and Product with New Leadership Structure

Hassabis will shift toward high-level strategy, advising on artificial general intelligence frameworks and expanding his role at Isomorphic Labs, Alphabet's drug discovery spinout. In a memo to staff, Hassabis said handing over operational responsibilities will allow him to concentrate on broader strategic work and biological applications of machine learning, particularly accelerated medical science and disease research.

The restructuring coincides with the departure of Jeff Dean after 27 years at Google. Dean, alongside Google Senior Fellow Sanjay Ghemawat, is leaving to launch an independent public benefit corporation focused on machine learning systems for science and engineering. Alphabet will serve as a founding investor and cloud infrastructure partner for the new entity, maintaining a joint research framework with both engineers.

The Cognarah Angle

This restructuring is Alphabet making a deliberate choice: commercial execution now comes first at Google DeepMind. The appointment of Kavukcuoglu to run operations, while Hassabis moves into an advisory position, officially divides the organisation into two tracks. One is focused on shipping software at scale today. The other is focused on long-term research that may or may not produce near-term revenue. That is not a subtle evolution. It is a structural admission that the two goals are no longer compatible under a single leadership model.

The pressure driving that choice is visible in the numbers. Nearly 950 million monthly users on the Gemini app and 900 million Gemma model downloads are metrics that demand the kind of cadence, reliability, and integration work that academic researchers are not hired to deliver. Alphabet is watching OpenAI and Microsoft move fast on enterprise contracts, developer tooling, and consumer adoption. Keeping a Nobel Prize-winning scientist in charge of quarterly shipping targets was never going to hold.

But the structural tension this creates is real. Fundamental research requires open-ended experimentation and tolerance for failure over long time horizons. Product organisations require roadmaps, deadlines, and predictable output. When operational control moves to a product-focused executive, the risk is not that research stops entirely. The risk is that it gradually bends toward whatever supports the next model release rather than whatever advances the science. Jeff Dean and Sanjay Ghemawat leaving to build an independent research vehicle, with Alphabet as a funder rather than an employer, may be the clearest signal yet of where that tension was already heading.

Alphabet may have built the most credentialed AI research organisation in the world. The question now is whether handing it a product manager's mandate will keep it that way.

Reporting sourced from Google Blog. Analysis and Cognarah Angle are Cognarah's own.