Anthropic's Earliest Backer Jaan Tallinn Believes AI Could End Humanity — And Still Expects to Make Billions From It

Jaan Tallinn, the Estonian co-creator of Skype and Kazaa, is on the verge of a multi-billion-dollar windfall from Anthropic's anticipated IPO — potentially valued above $2 trillion — while simultaneously warning anyone who will listen that artificial intelligence poses an existential threat to humanity. He spent the past week at the UN General Assembly making that case, and has been logging globe-spanning trips all year hitting AI summits, policymaker briefings, and ethical AI conferences. In February alone, he chartered a private jet from New Delhi to Paris that included AI researchers Stuart Russell and Yoshua Bengio, where the group debated machine consciousness over champagne at 40,000 feet.

Tallinn's wealth — estimated at around $12 billion, ranking him as Estonia's richest person — traces back to Skype, which Microsoft acquired for $8.5 billion, and early bets on AI companies including DeepMind (later bought by Google) and Anthropic. He led Anthropic's $124 million Series A in 2021 at a $461 million valuation, alongside Dustin Moskovitz, Eric Schmidt, and Jane Street's James McClave, at a time when most traditional venture capitalists passed, viewing Anthropic as a long shot against OpenAI. His logic for investing despite his fears: if he didn't, a safety-indifferent investor would have taken his place.

His journey into AI doom-thinking began in 2008, when he discovered the work of Eliezer Yudkowsky, the prolific rationalist writer and AI alignment advocate. After debating Yudkowsky, Tallinn became a major donor to safety organizations, eventually committing roughly $170 million to AI safety initiatives. He co-founded the Cambridge Centre for the Study of Existential Risk and the Future of Life Institute. In a 2011 phone call with Dario Amodei — who would later leave OpenAI to found Anthropic — Tallinn declared a double-digit probability that humanity dies from an AGI disaster, comparing civilization to people trying to defuse a live bomb. Amodei pushed back, suggesting that the actions meant to defuse the bomb might be the ones that trigger it.

Tallinn's actual influence inside Anthropic, however, has proven limited. He has acknowledged that as the company grew, individual investor voices became progressively quieter. His earlier experience at DeepMind was similarly humbling — he sat on its board but reportedly failed to move the founders on safety matters. Now, his investment firm Metaplanet is actively thinking about how to deploy anticipated IPO proceeds toward pro-safety causes. In perhaps the most striking illustration of Tallinn's worldview, his managing partner framed an Anthropic failure as a potential positive outcome: if the company goes to zero, it likely means artificial superintelligence wasn't reached, buying humanity extra time. As he put it, that's a win-win.