Anthropic just secured exclusive access to Colossus 1, SpaceX's massive supercomputer in Memphis, Tennessee, gaining 300 megawatts of computing power and over 220,000 graphics processing units (GPUs) within the month. The deal, announced on May 6, 2026, is remarkable not just for the sheer computing capacity involved, but because it represents a dramatic reversal between two companies whose leaders have spent months publicly feuding.

For Claude users, the practical impact is immediate. Anthropic announced three concrete improvements: Claude Code's five-hour rate limits are doubling for Pro, Max, Team, and Enterprise plans; peak-hours limit reductions are being removed for Pro and Max accounts; and API rate limits for Claude Opus models are being raised significantly. If you've watched Claude Code hit "you've hit your limit" walls during the workday, this deal is the infrastructure fix behind the scenes.

Anthropic

What Makes This Deal So Unusual?

The partnership is jarring because of the personal history between Elon Musk and Anthropic's leadership. In February 2026, Musk wrote on X that Anthropic "hates Western Civilization" and accused the company of being "doomed to become the opposite of its name." Then, abruptly, the tone shifted. In a post coinciding with the deal's announcement, Musk wrote that he had "spent a lot of time with senior members of the Anthropic team over the last week" and was "impressed".

Musk

"Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good," Musk stated.

Elon Musk, CEO of SpaceX and xAI

The reversal is awkward but strategically convenient. SpaceXAI is reportedly planning to go public as soon as next month, and locking in a marquee customer like Anthropic, a company in talks to raise at a $900 billion valuation, is exactly the kind of demand signal an initial public offering roadshow needs.

How Does This Fit Into Anthropic's Broader Infrastructure Strategy?

The Colossus 1 deal is not Anthropic's only major computing commitment. The company has stitched together a remarkably aggressive infrastructure buildout over the past several months:

• Amazon Partnership: An up to 5 gigawatt agreement, including nearly 1 gigawatt of new capacity by the end of 2026 using Trainium2 and Trainium3 chips
• Google and Broadcom Partnership: A 5 gigawatt agreement for next-generation TPU (Tensor Processing Unit) capacity beginning in 2027
• Microsoft and NVIDIA Partnership: A strategic partnership including $30 billion of Azure compute capacity
• Fluidstack Commitment: A $50 billion commitment to American AI infrastructure
• Google Cloud Services: A reported $200 billion commitment to Google's AI cloud services and TPUs, plus a further $40 billion Google investment in cash and compute

Most of those deals are forward-loaded, with capacity arriving in 2027 and beyond. The SpaceX agreement is different because it delivers immediate results. Three hundred megawatts within the month is not a roadmap promise; it's hardware already running in Tennessee. This makes Colossus 1 the bridge from the present compute crunch to future capacity.

Anthropic also continues to insist on hardware diversity. The company runs Claude on AWS Trainium chips, Google TPUs, and NVIDIA GPUs, a portfolio strategy that hedges against single-vendor risk and keeps negotiating leverage alive.

Why Is Anthropic So Desperate for More Computing Power?

Claude's surging popularity has created what Anthropic itself describes as "inevitable strain on our infrastructure," with reliability and performance suffering during peak hours. As reported by WIRED, the average developer is now spending at least 20 hours per week running Claude Code, a workload that has pushed the service into chronic congestion. The SpaceX deal is fundamentally an admission that the only way out is more silicon, and fast.

The scale of the Colossus 1 facility underscores just how much computing power is needed. Built from the ground up by xAI on a former Electrolux site in Memphis, SpaceXAI claims the supercomputer was stood up in 122 days. The facility houses dense deployments of NVIDIA's H100, H200, and next-generation GB200 accelerators, totaling roughly 220,000 GPUs. Anthropic isn't getting a slice of this capacity; it's getting the whole pie, effectively transferring a little under half of SpaceXAI's total 500,000-GPU fleet to Anthropic's workloads.

What Does This Mean for the Future of AI Infrastructure?

Perhaps the most provocative part of Anthropic's announcement is buried near the bottom: the company has "expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity". This is not entirely new ground for SpaceX. The company filed with the Federal Communications Commission in January to deploy a million-satellite orbital AI data center megaconstellation. SpaceXAI's argument is that "the compute required to train and operate the next generation of these systems is outpacing what terrestrial power, land, and cooling can deliver on the timelines that matter." Orbital compute, the company contends, offers "near-limitless sustainable power with less impact on Earth".

The engineering challenges are enormous: launch costs, radiation, heat dissipation in vacuum, latency, hardware reliability, orbital debris, and the practical problem of being unable to send a technician to swap a failed GPU. SpaceX's own filings reportedly acknowledge the venture involves "significant technical complexity and unproven technologies" and may never be commercially viable.

For now, orbital compute is more useful as an investor narrative than as a deployment plan. But the fact that a serious AI lab is willing to put the idea in writing tells you exactly how desperate the industry is for power that isn't gated by zoning boards and electrical substation queues.

How to Understand the Broader Compute Crisis in AI

• Power Constraints: Data centers require massive amounts of electricity, and many regions lack the power infrastructure to support new AI facilities, forcing companies to negotiate with utilities and governments for capacity
• Physical Space Limitations: Large-scale GPU clusters need enormous physical footprints, cooling systems, and real estate, making it difficult to expand quickly in populated areas
• Hardware Scarcity: NVIDIA GPUs are in high demand across the industry, and supply cannot keep pace with demand from AI labs racing to build larger models and serve more users
• Latency and Reliability: Moving compute to orbit introduces new technical risks, including signal delays and equipment failures that cannot be easily repaired

The Anthropic-SpaceX deal reflects a broader truth: the AI industry has hit a wall. Demand for computing power is colliding with the hard physical limits of power generation, land availability, and silicon production. Companies like Anthropic are no longer just building better models; they're scrambling to secure the raw computing infrastructure needed to keep their services running.