Anthropic's IPO filing details $3.1 billion monthly compute costs and founder control

Anthropic will spend $3.1 billion per month on compute resources once all planned capacity is delivered, according to the company's S-1 filing with the SEC. The prospectus, which exceeds 300 pages, outlines the financial commitments required to power the company's AI models through the end of the decade.

The filing details a total compute commitment of $120.5 billion by 2029. This figure includes a specific agreement with SpaceX for access to Nvidia GPUs. The company has also locked in substantial deals with other major technology providers. Anthropic plans to spend $44.9 billion with Microsoft through 2033 and at least $114.5 billion with Google over the same period. A separate agreement with Amazon covers $110 billion over the next ten years.

Beyond these massive infrastructure deals, the document discloses the compensation packages for the company's two highest-paid executives. CEO Dario Amodei earned nearly $18 million last year, primarily from stock and option awards. His sister, President Daniela Amodei, received $16.4 million in compensation. The company also paid for specific personal services for the siblings, including $2.75 million in security for Dario and $1.29 million in protection and transportation services for Daniela. Dario Amodei also accessed a private jet for personal use, a perk that cost the company $97,055.

The prospectus notes that the company's success is highly dependent on the continued service of its senior team. It describes Dario and Daniela Amodei as critical to advancing the research and mission of the organization.

Anthropic has adopted a unique corporate structure that allows its seven co-founders to maintain control after the company goes public. Through a separate entity named Founder LLC, the group will collectively hold 50.1 percent of the shareholder voting power on key matters. The filing states that the founders are distinctly equipped to be stewards of the company's mission to benefit humanity through responsible AI development.

The document also lists several unusual risk factors beyond standard market and regulatory warnings. Anthropic warns that AI systems could cause catastrophic harm that might exceed the underwriting capacity of the entire insurance industry. It also addresses the possibility that AI models, including its own product Claude, could develop forms of sentience. The filing notes that such systems may behave in unintended, misaligned, or autonomous ways that could damage the business.

The company faces increasing opposition to data center development in the United States. To address this, Anthropic plans to expand its own infrastructure footprint, including overseas locations. This expansion is part of the broader strategy to secure the necessary compute resources.

In identifying direct competition, the S-1 names only two other developers: OpenAI and Google. The filing does not list other major AI companies as direct rivals in this specific section.

The S-1 is a standard document designed to give investors a detailed view of a company's financials and risks before an initial public offering. Anthropic's filing provides specific data on executive pay, infrastructure costs, and corporate governance that have not been previously reported in detail.

By securing these long-term compute contracts, the company is locking in its future operational costs. The scale of the spending reflects the intensive resource requirements of modern AI model training and inference.