Google Deepmind is moving away from the dedicated research-team model that produced AlphaFold, one of its most visible scientific breakthroughs. The lab has broken up the group behind the protein-structure system, while several key researchers connected to the project are moving into other parts of Google or leaving for Anthropic.
What Changed Inside Google Deepmind
According to the Financial Times, Google Deepmind has broken up the team behind AlphaFold. Most of the original authors of the AlphaFold papers were reassigned internally over the past year, and Deepmind confirmed the moves.
The researchers are not all heading in one direction. Some are now working on projects built around Google's Gemini large language model. Others have moved into enzyme design, nuclear fusion, and genomics.
Some researchers also transferred to Isomorphic Labs, the Alphabet subsidiary focused on drug discovery. Isomorphic Labs was spun out of Deepmind in 2021 and has partnerships with Novartis and Eli Lilly.
The change is not only an internal reshuffle. Nearly a quarter of the full-time Google Deepmind authors of the original AlphaFold papers have left the company entirely, according to the FT.
Why AlphaFold Mattered
AlphaFold was developed starting in 2018. Its core role is to predict the three-dimensional structure of proteins, a problem that made the project a defining example of Deepmind's scientific ambitions.
The project also became closely tied to Deepmind's public identity. John Jumper, vice president and engineering fellow, and Deepmind CEO Demis Hassabis received the 2024 Nobel Prize in Chemistry for the work.
That recognition made AlphaFold more than another AI research project. It stood for a strategy: organize a dedicated team around a difficult scientific challenge, focus on a concrete goal, and pursue a result with potential global impact.
Pushmeet Kohli, vice president of research at Google Deepmind, described that earlier approach to the FT: "Our strategy over the last nine years has been to focus on grand challenges . . . a concrete goal every project is focused on," he said. "The strategy has evolved."
A Shift Toward Gemini-Powered Science
The new direction is broader than protein folding. Rather than organizing researchers only around single scientific problems, Deepmind is also building systems powered by Gemini that can assist scientists and eventually automate parts of the scientific process.
That means AlphaFold's legacy is being folded into a wider research and product agenda. The same organization that became known for a dedicated scientific breakthrough is now putting more attention on general systems that could support multiple areas of science.
The areas named in the reshuffle show that range clearly:
- Gemini large language model projects
- Enzyme design
- Nuclear fusion
- Genomics
- Drug discovery through Isomorphic Labs
This shift also comes as Deepmind competes with OpenAI and Anthropic to build frontier AI agents. In that context, scientific AI is no longer separate from the broader race to develop more capable AI systems.
Key Researchers Are Moving To Anthropic
The departures include researchers at the center of AlphaFold's success. Earlier this year, John Jumper and fellow AlphaFold researcher Jonas Adler moved to Google's internal "Code Strike" team, which was assembled to improve the company's AI coding capabilities as it seeks to catch Anthropic and OpenAI.
Last month, Jumper announced his move to Anthropic. He will be joined there by Adler and AlphaFold colleague Alexander Pritzel.
One Deepmind employee, who asked not to be named, told the FT that all three had been "instrumental, important, core members" of the company. The same employee said their departures had sparked surprise internally.
Anthropic has also moved into science-focused AI. The company launched Claude Science for biology and drug discovery earlier this year.
What The Restructuring Signals
Deepmind said it is "incredibly proud" of its scientific heritage and the global impact of AlphaFold, "which inspired the launch of Isomorphic Labs with the mission to reimagine drug discovery." That statement keeps AlphaFold positioned as a major achievement, even as the original team structure changes.
The larger signal is strategic. Deepmind is not abandoning science as a category. It is changing how scientific work is organized, moving from one dedicated AlphaFold-style group toward a wider mix of Gemini-powered systems, AI for science projects, and frontier AI agent development.
For Google Deepmind, that creates a different kind of challenge. AlphaFold showed what a focused research effort could accomplish. The next phase asks whether a broader AI platform strategy can produce scientific tools with comparable impact while the company also competes for talent with Anthropic and OpenAI.