en.Wedoany.com Reported - Elon Musk publicly corrected his previous assessment of AI company Anthropic, stating that Anthropic has become a key customer of SpaceX. In July 2026, Musk acknowledged on X that his September 2025 post claiming "Victory was never among the possible outcomes for Anthropic" was wrong, and noted that Anthropic is a current leader in the AI field, with its released Mythos/Fable model being one of the best in the industry.

Anthropic signed a long-term computing power lease agreement with SpaceX's xAI division. Under the agreement, Anthropic purchases all 300 megawatts of computing capacity from the Colossus 1 data center near Memphis, Tennessee, paying $1.25 billion per month from before May 2029, with a total contract value of approximately $40 billion. Additionally, Google also signed an agreement to lease SpaceX infrastructure, paying $920 million per month until June 2029.

Musk emphasized that he would not use infrastructure partnerships to squeeze out competitors, citing examples such as Tesla's 2014 patent pledge, opening the Supercharger network and charging port designs to competitors, and SpaceX launching competitive satellite systems under fair conditions. However, Musk also filed a lawsuit against OpenAI and acknowledged the existence of AI "distillation." Anthropic had accused three Chinese model manufacturers of distilling Claude in February.

Analysts believe that Anthropic does not need to worry about Musk unilaterally cutting off infrastructure, as contract terms and the substantial revenue SpaceX gains from the deal serve as constraints. Meanwhile, by hosting Anthropic's computing power, SpaceX may gain deep insights into the company's operations, but both parties are believed to have implemented security measures.