Full story
Colossus 1 Compute Deal
Anthropic and SpaceXAI have signed an agreement for Anthropic to use all of the compute capacity at xAI’s Colossus 1 data center in Memphis, Tennessee, giving Anthropic access to more than 300 megawatts of additional capacity (over 220,000 NVIDIA GPUs) within the month. The deal is aimed at improving capacity for Claude Pro and Claude Max subscribers, and Anthropic said it has “expressed interest” in partnering with SpaceX to develop “multiple gigawatts of orbital AI compute capacity.” In a post on X, Claude wrote, “We've agreed to a partnership with @SpaceX that will substantially increase our compute capacity.” Data Center Dynamics also described the partnership as one where Anthropic plans to use “all of the compute capacity at xAI's Colossus 1 data center.”
“We've agreed to a partnership with @SpaceX that will substantially increase our compute capacity”
Usage Limits and Musk
Alongside the compute access, Anthropic said it is doubling Claude Code’s five-hour rate limits for Pro, Max, Team, and seat-based Enterprise plans and removing the peak hours limit reduction on Claude Code for Pro and Max accounts. The company also said it is raising its API rate limits considerably for Claude Opus models, and Axios reported that Anthropic’s head of product Ami Vora said, “We're partnering with SpaceX to use all the capacity of their Colossus One data center.” Wired reported that Anthropic executives made the announcement on stage at the company’s annual developer conference in San Francisco, and it said SpaceXAI put out a blog post detailing that Anthropic would draw power from xAI’s Colossus 1 supercomputer. Musk’s posture toward Anthropic shifted as well: CNBC quoted him writing on X that he “spent a lot of time last week with senior members of the Anthropic team” and was “impressed.”
Capacity Race and Stakes
The partnership arrives as Anthropic has been dealing with demand that led to rate caps and usage limits, with Wired noting that software developers had complained about rate limits and service disruptions to Claude Code due to physical limits of available computing resources. Wired also reported that Anthropic said the average developer is now spending at least 20 hours per week running Claude Code, and it tied the compute deal to the broader scramble for enough computers to run complex AI software. SpaceXAI’s pitch for future capacity includes orbital development, with Wired stating that SpaceXAI said Anthropic has “expressed interest” in partnering to develop “orbital AI compute capacity.” The stakes extend to SpaceXAI’s effort to go public, since Wired said the partnership comes at a pivotal time for SpaceXAI, which is seeking to go public as soon as next month.
“expressed interest” in partnering to develop “orbital AI compute capacity”
