Anthropic announced on May 6 that it has secured access to the full compute capacity of SpaceX’s Colossus 1 data center in Memphis, Tennessee. The facility houses more than 220,000 Nvidia processors and will deliver 300 megawatts of new capacity to Anthropic within a month — one of the largest single compute expansions any AI company has announced this year.

The deal is notable for two reasons. First, it is one of the most significant infrastructure moves in the current AI compute race. Second, it signals a surprise shift in the relationship between Elon Musk and Anthropic, after Musk spent much of early 2026 publicly attacking the company.

The Compute Arms Race Has a New Entrant

Anthropic has been capacity-constrained. Claude users on paid plans have hit rate limits regularly, and the company has been working to expand its infrastructure access through deals with Amazon (AWS), Google Cloud, and Coreweave. The SpaceX Colossus deal is different in scale: 220,000 Nvidia processors at a single site, all going to Anthropic.

For context, large enterprise deployments of Claude have historically run into five-hour usage caps under peak demand. Anthropic has confirmed the Colossus capacity will be used to lift those caps immediately for subscribers on Claude Pro, Max, Team, and Enterprise plans.

That is a direct, practical change for businesses using Claude as a core part of their operations.

A Surprising Partnership

The partnership is politically awkward. In February 2026, Musk posted on X that Anthropic “hates Western civilization.” SpaceX and xAI had merged earlier this year, making Musk’s infrastructure organization and AI company one entity.

But after Musk spent time with senior members of the Anthropic team in late April, his public stance shifted. He described being “impressed” with the team and announced the compute deal shortly after.

It raises a straightforward question for businesses: does the source of compute matter if the result is faster, more reliable AI? In practice, most enterprise teams will not care whether Claude’s compute runs on SpaceX hardware or Google TPUs. What they care about is uptime, speed, and capability. By that measure, this deal is a net positive.

The Space Compute Angle

Beyond Colossus 1, Anthropic has expressed interest in working with SpaceX to develop compute capacity in orbit. SpaceX has been developing satellite-based infrastructure through Starlink, and the potential to run AI inference closer to the edge — or even in space — is early-stage but not theoretical.

This is a longer time horizon story. For 2026, the Colossus deal is what matters operationally. The space angle is worth tracking if you are thinking about where AI infrastructure heads over the next five to ten years.

What This Means for Business

Capacity directly affects usability. If your team is building workflows on Claude Enterprise and hitting rate limits, the Colossus expansion changes your operational ceiling. More capacity means more concurrent usage, longer sessions, and more headroom for agentic workloads that run autonomously over extended periods.

The compute race is accelerating. Google, Microsoft, Amazon, and now SpaceX are all competing to supply compute to frontier AI labs. That competition benefits enterprise buyers: it drives pricing pressure down and reliability up over time. Anthropic locking in 300 megawatts of new capacity strengthens its ability to compete with OpenAI and Google Gemini on availability.

AI reliability is becoming infrastructure-grade. JPMorgan Chase recently reclassified AI spending as core infrastructure, alongside data centers and payment systems. When businesses start treating AI that way, uptime and throughput matter as much as capability. Deals like this are what makes that reclassification credible.

The IPO signal. Coindesk reported that this deal is happening ahead of a reported Anthropic IPO in June. A company that needs to demonstrate operational scale to public investors is going to secure compute aggressively. For enterprise buyers, that typically means better enterprise support, clearer SLAs, and more investment in reliability over the next 12 months.

---

For businesses running AI workflows on Claude, the practical outcome is straightforward: rate limits are going up, capacity is expanding, and Anthropic is investing in the infrastructure to make enterprise-grade Claude use more reliable. That is the signal worth acting on. The Musk drama is a footnote.

If you are evaluating AI infrastructure for your business, the Omni Advisory service helps companies make sense of the AI vendor landscape — including which models and compute arrangements actually fit your operational needs.