I can't believe that the entire company's strategy is to disband the NOA team!
Google DeepMind has reorganized the core team that created the Nobel Prize-winning AlphaFold: no longer retaining an independent AlphaFold team, but distributing relevant researchers to multiple strategic projects. This adjustment is seen as an important signal that DeepMind will shift its focus further to Gemini's big model and AI Agent.
Most of the original AlphaFold team has been reassigned to Gemini, AI programming, genomics, enzyme design, nuclear fusion, and other internal projects, with some members transferred to Google's Isomorphic Labs, a biotechnology company.
Several key members, including Nobel laureate in chemistry John Jumper, researcher Jonas Adler and Alexander Pritzel, have joined Anthropic. Before leaving DeepMind, Jumper and Adler briefly joined the Code Strike team responsible for improving Google's AI programming capabilities. The team was formed to improve the company's AI coding capabilities as Google was trying to catch up with Anthropic and OpenAI. An anonymous DeepMind employee said the three were "critical, lighthearted core members" of the company, and their departure shocked the insiders.
The report also noted that nearly a quarter of the authors of the original AlphaFold paper have left DeepMind.
This change in personnel has also left the outside world wondering: How long will Hazabis be able to stay at Google?
DeepMind began the development of AlphaFold in 2018, which radically changed biological research by using the three-dimensional structure of AI-predicted proteins. This breakthrough won the 2024 Nobel Prize in Chemistry to John Jumper, vice president of engineering and DeepMind's chief executive officer. The project became a landmark example of DeepMind's long-term scientific breakthrough strategy through a dedicated research team.
At the same time, the project is also one of the model open source projects that Google DeepMind prides itself on.
And just hours before the Financial Times broke the news, Google's DeepMind had just updated AlphaFold v3.0.4.
Address of the project:https://github.com/google-deepmind/alphafold3
For this adjustment, DeepMind said that the company's strategy has evolved from "solving a single scientific problem" to "building a universal AI system at the core of Gemini that can fully accelerate scientific discovery". In the future, Gemini will become an important foundation platform for driving scientific innovation, rather than just serving a specific research project.
The analysis suggests that this change reflects a shift in the current competitive landscape of the AI industry. In the context of companies such as OpenAI, Anthropic and others continuing to encode their flagship big language models and AI Agent, Google is also investing more resources in building general AI capabilities to maintain competitive advantage.
It's worth noting that more than six months ago, DeepMind published a blog that summarized AlphaFold's five-year achievements and called it "the engine of global scientific discovery".
Link to the blog:
According to the blog, since the AlphaFold Protein Database went online in 2021, the database has collected more than 200 million protein structure predictions used by more than 3 million researchers in more than 190 countries and regions around the world, with more than a third of the related studies focusing on disease mechanics exploration.
According to official data, AlphaFold has been cited in more than 35,000 papers, and more than 200,000 others have incorporated it into research methods, and has led to breakthroughs in many fields such as drug development, cardiovascular disease research, and agricultural breeding.
DeepMind also stressed at the time that AlphaFold not only gave birth to Isomorphic Labs, but further developed a series of AI for Science models such as AlphaFold 3, AlphaGenome, AlphaProteo, and hoped to use AlphaFold as a template to apply AI to broader scientific research and accelerate scientific discovery.
However, only half a year later, when the AlphaFold independent team was withdrawn, core members resigned, and DeepMind's resources began to shift entirely to Gemini and AI Agent, the outside world began to re-examine whether the AI lab, once renowned for its basic scientific research, was moving toward a path of development centered on general AI products.
The link is as follows: