Anthropic is deepening its relationship with SpaceX, expanding its use of the company’s Colossus AI infrastructure while paying an astonishing US$15 billion per year for compute capacity.
The update came from today’s SpaceX S-1 filing and expanded upon by Anthropic co-founder Tom Brown, who revealed on X that the company will begin scaling up on NVIDIA GB200 capacity at Colossus 2 throughout June. Brown thanked Elon Musk and the broader SpaceX team “for helping us find good homes for the Claudes,” a reference to Anthropic’s Claude family of AI models.
SpaceX Moves Into AI Compute Services
The expansion marks another major step in SpaceX’s unexpected emergence as a large-scale AI infrastructure provider. While the company is best known for rockets, satellites, and Starlink, Musk confirmed that SpaceX is now actively offering AI compute as a service to outside companies at “significant scale.”
Musk also hinted that Anthropic may be only the beginning.
“As the recently expanded partnership with AnthropicAI demonstrates, SpaceX is offering AI compute as a service at significant scale,” Musk wrote on X. “We are in discussions with other companies to do the same.”
He added that SpaceX ultimately expects to provide AI services “at extremely high scale,” particularly through the use of orbital data centers, an idea Musk has referenced before as a way to overcome power and cooling limitations facing terrestrial AI infrastructure.
A Big Partnership
However, one of the biggest takeaways from the partnership may be the sheer amount of money involved.
Based on figures included in SpaceX’s IPO filing, Anthropic will be paying US$15 billion per year for compute capacity. That works out to $1.25 billion per month, or more than US$41 million every single day flowing to SpaceX for AI infrastructure access.
The massive payment highlights just how valuable large-scale GPU clusters and power capacity have become in the rapidly escalating AI arms race. Companies developing advanced AI models are now spending unprecedented amounts of money to secure access to next-generation hardware capable of training and operating increasingly complex systems.
Colossus 2 Expansion Already Underway
The latest announcement builds on the partnership first revealed earlier this month, when Anthropic secured access to more than 300 megawatts (MW) of computing capacity at Colossus 1. At the time, reports indicated Anthropic would consume the full capacity of the facility to support the growing demand for its Claude Pro and Claude Max AI services.
Now, with the addition of GB200 hardware at Colossus 2, the relationship appears to be expanding rapidly. NVIDIA’s GB200 platform is one of the most sought-after AI compute systems currently available, designed specifically for training and operating next-generation large language models and reasoning systems.
Always get the latest - make Drive Tesla your preferred source on Google
Are you buying a Tesla? If you enjoy our content and we helped in your decision, use our referral link to get three months of Full Self-Driving (FSD).Drive Tesla uses affiliate links, which means we may earn a commission if you make a purchase through them, at no additional cost to you. Thank you for your support.
