SpaceX to Power Anthropic’s Claude With Colossus 1

SpaceX will give Anthropic access to Colossus 1, adding more than 300 megawatts of compute capacity to support Claude Pro and Claude Max, the companies announced Wednesday.

Colossus 1 is the large AI supercomputer cluster developed under the xAI program to train and run advanced models. SpaceX merged with xAI in February and consolidated parts of the business under the name SpaceXAI. Company materials note that training work has moved to Colossus 2, a larger cluster intended to exceed Colossus 1’s roughly 220,000 Nvidia GPUs.

Anthropic co-founder Tom Brown wrote on X that the company will begin ramping Claude inference on Colossus in the coming days and thanked SpaceX for the partnership. He added, “We are going to need to move a lot of atoms in order to keep up with AI demand, and there’s nobody better at quickly moving atoms (on or off planet Earth).”

Elon Musk wrote on X that meetings with senior Anthropic staff led him to conclude the company approaches AI development responsibly. He wrote, “Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good.”

The agreement links two companies that have taken different positions on AI policy. Musk has aligned with former President Donald Trump and with the Departments of Justice and Defense on some policy questions. Anthropic has advocated stronger AI safety standards and declined government requests to permit unrestricted military use of Claude, citing safeguards against mass domestic surveillance and fully autonomous weapons. After those disputes, the administration directed federal agencies to phase out Anthropic’s technology while Pentagon officials emphasized contractors must allow “any lawful use.”

Anthropic said the Colossus capacity will improve reliability and performance for Claude Pro and Claude Max subscribers and that the companies plan to start shifting inference workloads to the SpaceX-run cluster in the near term.

Anthropic listed existing infrastructure partners that include Amazon, Google, Broadcom, Microsoft and Nvidia, and said it is interested in working with SpaceX to develop multiple gigawatts of orbital AI compute capacity. SpaceX and Anthropic noted limits on power, cooling and land could constrain future on-Earth data centers and that orbit-based compute could offer larger and more sustainable energy supplies if engineering challenges are addressed.

Separately, Musk acknowledged in court testimony that xAI partly used OpenAI models while training its Grok chatbot. The statement came as a federal trial opened this week to examine OpenAI’s governance and broader industry practices.