Anthropic Says It Will Use Full Capacity at SpaceX’s Colossus 1 Data Center

According to the new report, Anthropic’s deal with SpaceX provides access to 300 megawatts of compute power at the Colossus 1 data center in Memphis, Tennessee, to expand capacity for Claude models.

Fact Check
All key elements of the claim are directly confirmed by primary sources. Anthropic's own official blog post ('Higher usage limits for Claude and a compute deal with SpaceX') explicitly states the agreement to use all of SpaceX's Colossus 1 compute capacity, providing 300+ megawatts and 220,000+ NVIDIA GPUs, and confirms the intent to raise Claude usage limits and API rate caps. SpaceX's official announcement at x.ai/news/anthropic-compute-partnership corroborates the deal from the counterparty's perspective. CoinDesk and CNBC provide independent news corroboration. The only minor discrepancy is the date: sources consistently report May 6, 2026, while the claim title references May 7 - this is likely a timezone or publication-lag artifact and does not affect the substance of the claim.
Summary

Anthropic struck a deal with Elon Musk-owned SpaceX to expand computing capacity for its Claude AI models, with access to 300 megawatts of new compute power from SpaceX’s Colossus 1 data center in Memphis, Tennessee. The update reinforces that the partnership is centered on scaling Claude-related infrastructure and adds the location detail for the facility. Existing reporting on the topic said Anthropic would use all computing resources at Colossus 1, with more than 300 megawatts of additional deployed capacity expected within one month to raise limits for Claude Code and the Claude API.

Terms & Concepts
  • Colossus 1: A SpaceX data center facility that Anthropic said will provide its full computing resources under the partnership.
  • Claude API: Anthropic’s application programming interface that lets developers access Claude models programmatically.
  • Claude Code: A Claude product referenced by Anthropic as a target for higher usage limits under the added compute capacity.