Elon Musk Teams Up With Anthropic, a Company He’s Called ‘Evil’
SpaceX gives Anthropic access to 220,000-GPU Colossus data center after Musk's about-face.
Elon Musk's SpaceX has struck a surprising deal with Anthropic, the AI company behind Claude that Musk previously labeled 'misanthropic,' 'evil,' and accused of hating Western civilization. Under the agreement, Anthropic gains access to the massive Colossus data center, which houses 220,000 NVIDIA GPUs and has a 300-megawatt energy capacity. In return, Anthropic will double Claude Code's five-hour rate limits for Pro, Max, Team, and Enterprise plans, remove peak-hours limits for Pro and Max accounts, and increase API rate limits for Claude Opus. The move comes as SpaceX reportedly prepares for an IPO and needs to bolster its books, while its AI subsidiary xAI—now being absorbed into SpaceXAI—has not generated enough demand for the compute power it hastily built at Colossus.
Musk's change of heart appears driven by personal interactions: he claimed to have 'spent a lot of time last week with senior members of the Anthropic team' and came away 'impressed,' adding that 'no one set off my evil detector.' This odd partnership underscores the fluid alliances in the AI industry, even amid bitter rivalries. Critics note the irony of SpaceX, which sends astronauts into space, now being the same company behind a 'pornographic chatbot' (referring to Grok's lax content filters). The dissolution of xAI and the Colossus deal mark a new chapter for Musk's AI ambitions, blending compute leasing with his space ventures.
- SpaceX provides Anthropic access to Colossus data center with 220,000 NVIDIA GPUs and 300 MW capacity.
- Anthropic raises Claude Code rate limits for Pro/Max/Team/Enterprise plans and expands API access to Opus.
- Musk reverses prior criticism of Anthropic after meetings, while dissolving xAI into SpaceXAI ahead of IPO.
Why It Matters
This partnership may reshape AI compute leasing and signals that even bitter rivals will collaborate for resources.