Overview
- The agreement announced Wednesday gives Anthropic use of SpaceX’s Memphis‑based Colossus 1, which packs more than 220,000 Nvidia GPUs and over 300 megawatts of capacity.
- Anthropic said it will begin shifting Claude inference to the cluster within days to cut slowdowns and reduce outages during peak use.
- Following the announcement at its Code with Claude event, the company doubled five‑hour rate limits for Pro and Max, removed peak‑hour caps for Claude Code, and raised API quotas for Opus models.
- Both companies said they are exploring space‑based data centers that could deliver multi‑gigawatt AI compute, describing the effort as an early collaboration.
- SpaceX confirmed xAI is now folded into “SpaceXAI,” and Elon Musk softened his stance on Anthropic after meeting its leaders, as the ready‑to‑run Colossus capacity fills a gap before Anthropic’s larger cloud deals arrive in late 2026.