Google DeepMind has dismantled the team behind its Nobel prize-winning AI system that predicts protein structures in a major shift of its research direction, the Financial Times has reported.

Citing people familiar with the matter and its own analysis of recent job moves, the FT has reported that the majority of the original authors of the AlphaFold papers have been reassigned over the past year.

DeepMind confirmed to the paper that the staff have moved internally to projects built around Google’s large language model, Gemini, as well as areas including enzyme design, nuclear fusion and genomics.

Others have relocated to Isomorphic Labs, a drug discovery company owned by Google’s parent, Alphabet. Nearly a quarter of the full-time Google DeepMind authors of the papers have left the company altogether.

“Our strategy over the last nine years has been to focus on grand challenges […] a concrete goal every project is focused on,” Pushmeet Kohli, vice-president of research at Google DeepMind, told the FT. “The strategy has evolved.”

Rather than only organising scientists around single scientific problems such as protein folding, Kohli told the paper that DeepMind is now also focused on building Gemini-powered systems that can assist scientists while also competing with OpenAI and Anthropic to develop frontier models.

AlphaFold, which began development in 2018, uses AI to predict the three-dimensional structure of proteins, transforming biological research. The technology earned chemist and DeepMind director John Jumper and chief executive Demis Hassabis part of a shared Nobel Prize in chemistry in 2024.

---