Google DeepMind disbands its Nobel-winning AlphaFold team, reassigning researchers to Gemini-centred AI as key authors leave for Anthropic

Summary

Google DeepMind has dissolved the dedicated team behind AlphaFold, the AI system that earned its lead researchers the 2024 Nobel Prize in Chemistry for predicting protein structures from amino acid sequences. Most researchers have been reassigned to a Gemini-centred scientific AI programme within Google DeepMind. Chief AlphaFold author John Jumper and at least two colleagues have left the company, with some joining Anthropic. Nearly a quarter of the original team has left DeepMind entirely. The move, first reported by the Financial Times, signals a strategic pivot from standalone domain-specific scientific AI toward general-purpose foundation models anchored to the Gemini family.

The split

Engadget and PYMNTS led with the blunt news: the Nobel project is over. The Decoder, reporting from Germany, was more granular, emphasising the scale of the talent departure, specifically that almost a quarter of the team has left DeepMind, and framing the restructuring as a departure from the scientific-AI identity that distinguished DeepMind from other AI labs. XenoSpectrum, citing the Financial Times, contextualised the move within the competitive landscape: protein-structure prediction has spread across the full research pipeline since AlphaFold's peak, reducing the strategic value of maintaining a dedicated team. No outlet critical of the disbandment, or representing a scientific-community perspective on the loss of the AlphaFold unit, appeared in this feed.

By the numbers

• 2024, the year AlphaFold's researchers received the Nobel Prize in Chemistry
• 25 percent (approximately), the share of the AlphaFold team that has left DeepMind entirely
• 2, key AlphaFold co-authors confirmed to have left for Anthropic alongside John Jumper

Why it matters

AlphaFold was the clearest demonstration that AI could solve a problem in structural biology that had defeated human researchers for decades. Its dissolution as a standalone project suggests DeepMind now treats scientific breakthroughs as proofs of concept for general-purpose foundation models, not as sustained research programmes. The talent migration to Anthropic, a safety-focused frontier lab, adds to the broader picture of experienced AI researchers reorienting around safety concerns, a shift also visible in the AI pacing petition signed by staff at both companies on July 28.

What to watch

• Whether the Gemini-centred scientific AI programme produces outcomes comparable to AlphaFold, or whether folding domain-specific AI into a general foundation model reduces research impact
• The trajectory of AlphaFold's successor tools, which other labs and startups have been building since AlphaFold 2 made protein-structure prediction broadly accessible
• Any statement from John Jumper or other departed researchers about their scientific agenda at Anthropic
• How Google DeepMind frames its scientific mission now that its most publicly celebrated project no longer has a dedicated team