Anthropic has entered into a substantial computing agreement with SpaceX, securing full access to the Colossus 1 supercomputer located in Memphis, Tennessee. The deal, announced on May 6, 2026, provides the artificial intelligence company with over 300 megawatts of compute capacity, equating to more than 220,000 Nvidia GPUs. This influx of processing power is expected to come online within the month and will directly improve capacity for Anthropic's Claude Pro and Claude Max subscribers. The partnership comes at a critical time for Anthropic, which has been seeking to expand its compute resources. While Anthropic has existing compute agreements with major cloud providers like Amazon, Google, and Microsoft, the capacity from those arrangements is not anticipated to be fully available until late 2026 or early 2027. The SpaceX deal offers immediate relief, allowing Anthropic to increase usage limits for its Claude Code product and its Claude Opus models. Specifically, Anthropic is doubling the five-hour rate limits for Claude Code for paid users and removing peak-hour restrictions for Pro and Max users. API rate limits for Claude Opus models are also being considerably increased. SpaceX acquired xAI, the entity that built the Colossus 1 supercomputer, earlier this year. The Colossus 1 facility features a dense deployment of Nvidia's H100, H200, and next-generation GB200 accelerators. SpaceXAI, the AI division of SpaceX, stated that Colossus 1 was built in record time and delivers significant scale for AI training, fine-tuning, inference, and high-performance computing workloads. Reports indicate that xAI's own utilization of Colossus 1 had been low, around 11%, prompting the company to shift its training operations to its Colossus 2 supercomputer. This situation positions SpaceX to monetize an underutilized asset. Beyond terrestrial computing, the agreement also includes an expression of interest from Anthropic to partner with SpaceX on developing multiple gigawatts of orbital AI compute capacity. SpaceX has previously filed paperwork with the Federal Communications Commission indicating its intent to launch a large satellite constellation for orbital data centers. The companies suggest that space-based compute could offer near-limitless sustainable power with reduced environmental impact on Earth, addressing the growing demand for AI processing that is outpacing terrestrial power and cooling capabilities. Elon Musk, CEO of SpaceX, has previously stated that orbital data centers could become the most cost-effective method for increasing compute power within two to three years. The partnership occurs despite past public criticism from Elon Musk directed at Anthropic. Musk had previously labeled Anthropic as "misanthropic" and "evil." However, Musk stated that after meeting with senior members of the Anthropic team, he was "impressed" and that "no one set off my evil detector." He added that the team appeared "highly competent and cared a great deal about doing the right thing," and that Claude would "probably" be good as long as the company engaged in "critical self-examination." Musk also reserved the right to reclaim the compute capacity if Anthropic's AI engages in actions that harm humanity. This deal diversifies Anthropic's compute supply chain, which already includes agreements with Amazon Web Services, Google, and Microsoft. For SpaceX, the partnership signals its move toward becoming a significant AI infrastructure provider, leveraging its acquisition of xAI and its capabilities in large-scale deployment and operations.