For years, the founding story of Anthropic has been told as a split within OpenAI: a group of researchers led by Dario Amodei left in 2020 to build a rival lab dedicated to safer AI. The list of early supporters, however, included a name that has never been made public until now: Demis Hassabis, the Google DeepMind chief executive who leads the other half of the industry’s most tangled rivalry.
Hassabis, a Nobel laureate, was an angel investor in Anthropic, according to people familiar with the matter. The stake, previously undisclosed, adds another layer to the relationship between Google and a startup that has become its most formidable challenger in frontier AI. Google has separately poured billions of dollars into Anthropic through its cloud division and an artificial-intelligence partnership, people familiar with the arrangements said.
A person close to Anthropic said its chief executive, Dario Amodei, has long treated Hassabis as a role model in the field. Investors now value the company at around $900 billion, according to people familiar with its latest funding talks.
Hassabis’s personal backing of Anthropic is striking for a simple reason: he runs Google’s flagship AI laboratory, and Anthropic’s Claude models compete head-on with Google’s Gemini line. Yet the investment fits a pattern that predates the current AI boom. In 2014, Hassabis sold his London lab, DeepMind, to Google for £400 million. Since then, he has repeatedly appeared as an angel investor in companies founded by former colleagues and students.
The scale of that diaspora has grown quickly. Researchers who left DeepMind since 2021 have founded more than a dozen AI startups that have raised at least $14 billion in total, according to data from PitchBook and Dealroom. Two of them, Chiral, a London-based lab that Hassabis also backs, and Nanyan Intelligence, raised more than $3 billion combined in the past month alone.
The disclosures land at a delicate moment for both companies. Google is racing to defend its position in AI search and consumer products while pouring resources into Gemini. Anthropic is scaling fast, signing enterprise contracts and pushing Claude into developer tools that overlap with Google’s own offerings. The two companies compete in cloud credits, model quality, and recruiting, yet the web of personal and financial ties keeps pulling them together.
Hassabis’s angel investing has been a quiet counterpoint to his public role. The Nobel laureate in chemistry, honored for work on protein structure, has spent much of the past two years on the road arguing that AI needs careful stewardship. His personal portfolio tells a different story: a man who keeps betting, with his own money, on the people he once trained.
Whether his stake in Anthropic was made before the company’s valuation run-up, and how large it is, could not be learned. People familiar with the matter declined to discuss the size of the position, and Anthropic’s cap table has become one of the most scrutinized in the industry as investors circle a potential public listing.
The revelation also complicates the narrative around Google’s relationship with Anthropic. Regulators on both sides of the Atlantic have been examining the company’s multibillion-dollar cloud deals with the startup, and a personal investment by Google DeepMind’s chief is the sort of detail that antitrust lawyers tend to notice. Neither Google nor Anthropic responded to requests for comment.
For Amodei, Hassabis’s endorsement carries a symbolic value that money alone cannot buy. The younger executive has spoken publicly about the debt his field owes to DeepMind’s early work, and colleagues say he has cited Hassabis’s career as a template: build a lab, prove the science, then shape the industry’s rules.
Hassabis has not hidden his doubts about parts of the industry’s direction, and the investment sits awkwardly next to some of his public statements. He has told interviewers that the race to build general-purpose AI is moving faster than he would like and that commercial pressures distort the field’s priorities. People who know him said the Anthropic stake reflects a different calculation: a belief that the startup’s cautious approach to deployment is the right model for the industry’s next phase.
The two men at the center of the story share a style as much as a field. Amodei, like Hassabis, came up through academic research, and he has said repeatedly that Anthropic exists to make sure powerful AI remains controllable. His investors are betting that careful scaling wins the long game. Hassabis’s quiet check is a personal version of that same bet, and it will be parsed in Silicon Valley as the endorsement of the one person most entitled to resent Anthropic’s rise.
What remains unclear is how the relationship evolves as the two companies’ commercial interests collide. Google’s cloud division both funds Anthropic and competes with it; Hassabis’s personal stake now straddles the same line. If the history of AI’s founding generation is any guide, the ties will keep tightening, and the conflicts will keep multiplying.
