The artificial intelligence industry loves to narrate itself as a clash of titans—Google versus OpenAI versus Anthropic, billions of dollars and the future of humanity hanging in the balance. The reality, as it turns out, is cozier. Demis Hassabis, the Nobel Prize-winning co-founder of Google DeepMind, was an early investor in Anthropic, the safety-focused AI lab now valued at tens of billions and positioned as one of Google's primary competitors.
The revelation, reported by the Financial Times, punctures the tidy competitive framing that dominates coverage of the AI race. It also illuminates something more interesting: the extent to which a remarkably small network of researchers, funders, and true believers has shaped the entire trajectory of the technology.
The Hassabis network
Hassabis sold DeepMind to Google in 2014 for a reported £400 million, cementing his status as the field's most credentialed scientist-entrepreneur. His 2024 Nobel Prize in Chemistry—shared for work on protein structure prediction—only amplified his influence. Now his former colleagues, collaborators, and intellectual descendants occupy senior positions across every major AI organization. Dario and Daniela Amodei, who founded Anthropic after departing OpenAI, represent one branch of a sprawling tree that traces back, in various ways, to the ideas and institutions Hassabis helped build.
The investment itself is not illegal or even necessarily improper. Hassabis made the bet before Anthropic became a direct threat to Google's AI ambitions, and early-stage investing in promising startups is standard practice among wealthy technologists. But it does create an awkward geometry. Google has since invested billions in Anthropic—reportedly around $2 billion—while simultaneously competing with it for enterprise customers, developer mindshare, and the amorphous prize of "AGI." Hassabis, meanwhile, runs the division tasked with keeping Google ahead.
Why it matters beyond the gossip
The concentration of influence here is not merely a curiosity for industry watchers. It has implications for how AI safety is debated, how talent flows between organizations, and how seriously we should take the adversarial framing that companies use when lobbying regulators or courting investors. If the people building these systems share capital, intellectual lineage, and social ties, the "race" metaphor starts to look less like the Cold War and more like a gentleman's agreement with competitive flourishes.
This also raises questions about what "AI safety" means when its most prominent advocates are financially entangled with multiple outcomes. Anthropic has built its brand on being the responsible alternative to OpenAI's move-fast ethos. That positioning is harder to parse when the field's most decorated scientist has a stake in both the cautious approach and the aggressive one.
Our take
None of this makes Hassabis a villain or Anthropic a fraud. It makes them human, operating in an industry where the pool of qualified participants is small and the financial incentives are enormous. The real lesson is epistemic: when AI executives warn of existential risk or promise transformative benefit, remember that they are not disinterested observers. They are players in a game where the referees, commentators, and team owners often turn out to be the same few people.
