Anthropic S-1 Reveals Massive Compute Commitments and Founder Control
AI-generated image · US National Wire
The AI startup's IPO filing details hundreds of billions in spending with hyperscalers and a corporate structure designed to keep co-founders in power.
Anthropic has filed its S-1 prospectus with the SEC, disclosing the staggering scale of the compute investments required to sustain its AI models. As first reported by The Verge, the company has committed to spending at least $114.5 billion with Google through 2033, $110 billion with Amazon over the next decade, and $44.9 billion paid to Microsoft by 2033.
Further expanding its infrastructure, Anthropic has entered into an agreement with SpaceX for Nvidia GPU access. The Verge reports this commitment could reach $120.5 billion by 2029, with monthly costs hitting approximately $3.1 billion once full capacity is delivered. The company also noted plans to expand its infrastructure footprint internationally due to rising opposition to U.S. data center development.
Beyond capital expenditures, the filing reveals a corporate structure designed to preserve founder influence. Through an entity known as "Founder LLC," seven co-founders will maintain 50.1 percent of shareholder voting power on key matters. The filing identifies CEO Dario Amodei and President Daniela Amodei as critical to the company's mission; last year, Dario's compensation totaled nearly $18 million and Daniela's reached $16.4 million. The company also spent $2.75 million on personal security for Dario and $1.29 million for Daniela's security and transportation.
In its risk disclosures, Anthropic warns of "catastrophic" harms that could exceed the capacity of the entire insurance industry, as well as the possibility that its Claude AI could develop "forms of sentience." While the AI market is crowded, The Verge notes that Anthropic identifies only Google and OpenAI as its direct competition.