Anthropic and SpaceX Sign Compute Deal for 300MW Colossus 1 Capacity

Anthropic on May 6, 2026 signed an agreement with SpaceX that gives it the entirety of the compute capacity at Colossus 1, the hyperscale data center in Memphis, Tennessee built by xAI before its merger with SpaceX. According to Bloomberg, CNBC, and Anthropic's own announcement, the deal delivers more than 300 megawatts of new capacity and over 220,000 NVIDIA GPUs coming online within the month — one of the largest single compute commitments any AI lab has disclosed.

The capacity arrives at exactly the moment Anthropic's product limits had become a public pain point. The company moved immediately on usage: Claude Opus API rate limits went up, the five-hour rate limits on Claude Code doubled for Pro, Max, Team, and seat-based Enterprise plans, and the peak-hours reduction on Claude Code was removed for Pro and Max accounts. Anthropic also said it has "expressed interest" in working with SpaceX on multiple gigawatts of compute in space — a longer-term thread that turns the deal into more than a data-center lease.

The strategic context is unusually loaded. Colossus 1 originated as xAI's flagship training cluster; the xAI-SpaceX merger consolidated AI compute under the SpaceX umbrella, and now its largest tenant is a direct competitor to Elon Musk's own model line. For Anthropic, it diversifies away from Amazon and Google in the same quarter that Microsoft and OpenAI renegotiated their exclusivity arrangement — another sign that frontier labs are unbundling from single-cloud relationships. For SpaceX, the deal lands weeks ahead of its rumored June 2026 IPO and reframes the company as a serious AI-infrastructure supplier ahead of pricing.