Elon Musk's SpaceX and Anthropic have signed an unusual deal to give Anthropic access to the Colossus 1 AI supercomputer cluster, boosting capacity for Claude Pro and Claude Max subscribers. The agreement provides over 300 megawatts of compute power, with Anthropic planning to ramp up Claude inference on Colossus in the coming days.
Musk said conversations with senior Anthropic staff convinced him the company was approaching AI development responsibly, adding that he was impressed by their competence and commitment to doing the right thing. Meanwhile, SpaceXAI has already moved training to the larger Colossus 2 supercomputer cluster.
The partnership pairs two companies that have taken sharply different positions on AI regulation. Musk has aligned closely with Trump and the Pentagon, while Anthropic has refused to lift safeguards for unrestricted military use, leading to a federal order to phase out its technology.
Despite past tensions-Musk previously called Anthropic “misanthropic”-the deal highlights how the race for computing power is reshaping Silicon Valley alliances. Anthropic said it has expressed interest in partnering with SpaceX to develop orbital AI compute capacity, citing space-based compute as a potential source of sustainable energy.