ANTHROPIC COMMANDEERS ELON’S GPU ARMY — 220,000 NVIDIA CHIPS, ALL OF THEM, TO POWER CLAUDE The most unexpected partnership in AI this year just became official. Anthropic has signed an agreement with SpaceX to take over the full compute capacity of Colossus 1, Elon Musk’s AI training cluster in Memphis. That means 220,000-plus NVIDIA GPUs. That means 300 megawatts of processing power. That means Anthropic, locked in an existential race with OpenAI, just plugged itself into one of the most powerful AI computing facilities in the world outside of the hyperscalers. Let the awkwardness sit for a moment. Musk is currently appealing his failed court case against OpenAI. His xAI company competes directly with Anthropic. And his SpaceX subsidiary just handed its entire supercomputer to one of xAI’s most dangerous rivals. Business is business, apparently. The immediate effect for Claude users is real. Anthropic doubled rate limits for Claude Code across Pro, Max, Team, and seat-based Enterprise plans effective immediately. Peak-hour throttling on Pro and Max accounts is gone. API rate limits for Claude Opus models went up. The company said the extra capacity directly improves availability for subscribers who have been hitting limits during high-demand hours. Anthropicis not stopping there. The company also holds a five-gigawatt agreement with Amazon, a five-gigawatt partnership with Google and Broadcom coming online in 2027, a $30 billion Azure deal with Microsoft, and a $50 billion infrastructure commitment with Fluidstack. It has also expressed interest in developing orbital AI compute with SpaceX. The compute arms race is escalating faster than most people realize. Keywords: Anthropic SpaceX Colossus deal, Claude compute capacity, NVIDIA GPU infrastructure, xAI Anthropic compute