Anthropic Gets in Bed With SpaceX as the AI Race Turns Weird
Anthropic gains access to 220,000 Nvidia GPUs from Musk's SpaceXAI supercomputer.
Anthropic announced a partnership with SpaceXAI (the merged entity of SpaceX and xAI) to access computing resources from xAI's Colossus 1 supercomputer in Memphis. The 220,000-Nvidia-GPU facility, ranging from H100 to GB200 chips, was built in 122 days and is among the world's largest AI supercomputers. The deal gives Anthropic over 300 megawatts of new capacity, crucial for alleviating rate limits and disruptions on Claude Code as developer usage surges. Notably, Elon Musk, who previously criticized Anthropic's policies as 'misanthropic,' now praises the team's work after spending time with them.
The partnership boosts Anthropic's ability to serve Claude Pro and Claude Max subscribers while positioning SpaceXAI for its upcoming IPO by demonstrating customer demand. However, environmental protests have emerged over gas turbine emissions at the Memphis data center. Separately, Anthropic has committed $200 billion to Google's cloud and TPU services and $100 billion to Amazon's technologies over the next decade, highlighting the AI industry's insatiable compute appetite. Major cloud providers now count Anthropic and OpenAI contracts as over half of their $2 trillion backlogs.
- Partnership grants Anthropic access to 220,000 Nvidia GPUs and 300 MW of compute capacity from Colossus 1.
- Colossus 1 was built in 122 days; Musk previously criticized Anthropic but now praises their AI safety work.
- Capacity will reduce rate limits and disruptions for Claude Pro and Claude Max subscribers.
Why It Matters
Anthropic's deal with SpaceXAI shows AI labs' massive compute hunger, driving unusual partnerships and space-based data center ambitions.