Google DeepMind has dismantled the dedicated team behind AlphaFold, the Nobel Prize-winning protein-structure prediction system, reassigning most of its researchers to projects centered on the Gemini large language model and other scientific areas. The reorganization, confirmed on July 30, 2026, reflects a strategic pivot away from problem-specific teams toward broader AI systems for scientific discovery.
Most researchers who authored the original AlphaFold papers have been reassigned over the past year. Google DeepMind confirmed that employees have moved to projects involving Gemini, enzyme design, genomics and nuclear fusion. Some have joined Alphabet's drug discovery company, Isomorphic Labs, while nearly a quarter of the full-time authors of the original AlphaFold papers have left the company.
"Our strategy over the last nine years has been to focus on grand challenges... a concrete goal every project is focused on. The strategy has evolved," Pushmeet Kohli, Vice President of Research at Google DeepMind and founder of its AI for Science team, told the Financial Times.
According to Kohli, the company is no longer organizing researchers solely around individual scientific problems such as protein folding. Instead, it is developing Gemini-powered systems that can assist scientists and eventually automate parts of the scientific process, while competing with OpenAI and Anthropic to build frontier AI agents.
Strategic shift toward general-purpose scientific AI
AlphaFold, first developed in 2018, became one of DeepMind's defining achievements after using AI to predict the three-dimensional structure of proteins. The breakthrough earned DeepMind Chief Executive Demis Hassabis and Vice President John Jumper the 2024 Nobel Prize in Chemistry.
The restructuring aligns with a broader move across the AI industry, where leading labs are deploying large language models to accelerate scientific research. Anthropic has introduced Claude Science, a version of its AI assistant designed for biology and drug discovery, while competition for researchers capable of building frontier AI systems continues to intensify.
Personnel changes and departures
Earlier this year, Jumper and AlphaFold researcher Jonas Adler joined Google's internal Code Strike team to strengthen AI coding capabilities. Last month, Jumper announced he would leave for Anthropic, where Adler and fellow AlphaFold researcher Alexander Pritzel are also set to join. A DeepMind employee, speaking anonymously, said the three researchers had been among the company's core contributors and that their departures had surprised colleagues internally.
DeepMind said the movement of researchers reflects its evolving scientific priorities and highlighted AlphaFold's lasting impact, noting that the project led to the creation of Isomorphic Labs to commercialise its breakthroughs in drug discovery. The company spun out Isomorphic Labs in 2021, and the business has since signed partnerships with Novartis and Eli Lilly.
Why this matters for science and research professionals
The dismantling of the AlphaFold team signals a shift in how large AI labs are approaching scientific research. For professionals in science and research, the move underscores the growing importance of general-purpose AI systems like Gemini that can be applied across multiple domains. Staying current with these tools is becoming essential. The AI Learning Path for Research Scientists offers a structured way to build practical skills with AI systems now shaping the future of scientific discovery.
As labs like Google DeepMind and Anthropic compete to build AI agents that can automate parts of the scientific process, researchers who understand how to work with these models will be better positioned to lead. The field of AI for Science & Research is changing quickly, with new tools and methods emerging from the same forces that reshaped the AlphaFold team.
Your membership also unlocks:
