In February 2026, Elon Musk posted on X that Anthropic's models were "misanthropic and evil" and that the company hated Western civilization. About ninety days later, he handed them his supercomputer. Now it's in an SEC filing.

SpaceX filed its S-1 prospectus on May 20, confirming that Anthropic is paying $1.25 billion every month for AI compute access through May 2029. That works out to $41 million per day, $15 billion per year, and up to $45 billion over the full contract term. The filing targets the largest IPO in Wall Street history, with a planned Nasdaq listing under ticker SPCX as early as June 12.

Anthropic Could Post a $559 Million Operating Profit Years Before OpenAI Reaches Profitability
The numbers behind how Claude’s maker beat ChatGPT to profitability.

The Deal Musk Can't Explain With a Straight Face

SpaceX signed Cloud Services Agreements with Anthropic on May 6, giving the Claude maker full access to Colossus 1 and Colossus 2, the Memphis supercomputer campus housing 220,000 Nvidia GPUs and more than 300 megawatts of power.

Either party can walk away with 90 days notice, and the S-1 confirms SpaceX is already in talks with other AI companies for similar arrangements. "We expect to enter into additional similar services contracts," the filing states.

Yesterday on X, Musk explained the arrangement in his own words. "As the recently expanded partnership with Anthropic demonstrates, SpaceX is offering AI compute as a service at significant scale," he wrote.

Subscribe for free to continue reading this article

Subscribe Subscribe

Already Have an Account? Log In