SpaceX Gives Anthropic Access to Colossus 1

The Story

SpaceX has signed a deal to give Anthropic access to Colossus 1, the large AI supercomputer in Memphis, Tennessee, previously tied to xAI and now under the combined SpaceXAI structure. The agreement gives Anthropic access to more than 300 megawatts of new capacity and over 220,000 NVIDIA GPUs, with access expected within the month. Anthropic plans to use the extra compute to improve capacity for Claude subscribers and immediately raise usage limits across several products. The companies have also discussed a possible future partnership to develop multiple gigawatts of orbital AI computing capacity.

The Facts

• The deal gives Anthropic access to Colossus 1.

  Anthropic has signed an agreement with SpaceX to use the full compute capacity at the Colossus 1 data center.

• Colossus 1 is located in Memphis, Tennessee.

  The facility gives Anthropic access to more than 300 megawatts of new capacity and over 220,000 NVIDIA GPUs.

• The system uses high-end NVIDIA accelerators.

  Colossus 1 includes dense deployments of H100, H200 and GB200 chips.

• SpaceXAI built Colossus 1 on a former Electrolux site.

  The company says the supercomputer was built in 122 days and describes it as one of the largest and fastest AI supercomputers in the world.

• Claude Pro and Claude Max users are expected to benefit directly.

  Anthropic plans to use the additional compute to boost capacity for subscribers of its paid Claude AI assistants.

• Claude Code limits are being doubled.

  Anthropic is doubling Claude Code’s five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans.

• Peak-hours limit reductions are being removed for Pro and Max accounts.

  The change is designed to make Claude access more consistent during high-demand periods.

• Claude Opus API rate limits are increasing.

  Anthropic is also raising API rate limits for its Claude Opus models.

• Claude Code demand has strained available compute.

  Developers have complained in recent months about rate limits and service disruptions, while Anthropic says the average developer now spends at least 20 hours per week running Claude Code.

• The companies are discussing orbital AI computing.

  Anthropic has expressed interest in working with SpaceX to develop multiple gigawatts of AI compute capacity in orbit.

• Space-based compute remains unproven.

  SpaceX says orbital compute could use solar power directly in Earth’s orbit and reduce land and cooling constraints, but the concept still faces major technical and environmental hurdles.

• Colossus 1 has drawn environmental scrutiny.

  Residents and activists have raised pollution concerns around gas-burning turbines used to power the Memphis site.

How to Access / Pricing

Anthropic says the capacity increases are effective immediately across several products. The announced changes include doubled Claude Code five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans, removal of peak-hours limit reductions for Pro and Max accounts, and higher API rate limits for Claude Opus models.

Numbers that Matter

• 300+ megawatts of new capacity at Colossus 1.
• 220,000+ NVIDIA GPUs available through the Memphis facility.
• H100, H200 and GB200 NVIDIA accelerators deployed in the system.
• 122 days to build Colossus 1, based on SpaceXAI’s stated timeline.
• 20+ hours per week is how long the average developer now spends running Claude Code, based on Anthropic’s figures.
• More than $100 billion over ten years committed by Anthropic to Amazon technologies.
• $200 billion reportedly committed by Anthropic to Google AI cloud services and TPU chips.
• $30 billion of Azure capacity covered by Anthropic’s strategic partnership with Microsoft and NVIDIA.
• $50 billion infrastructure investment with Fluidstack.

Risks / Limitations

Colossus 1 has already faced pollution concerns from residents and activists in Memphis. Critics have pointed to gas-burning turbines brought in to supply power, saying the equipment has degraded air quality in surrounding neighborhoods.

The orbital AI compute idea is also still speculative. Space-based data centers could offer access to solar power and reduce some terrestrial constraints, but the concept faces unresolved technical and environmental challenges.

Background

Anthropic’s SpaceX agreement comes as AI companies race to secure enough compute to train and serve increasingly powerful models. The company has already announced major infrastructure deals with Amazon, Google, Broadcom, Microsoft, NVIDIA and Fluidstack, while also expanding some Amazon-backed capacity in Asia and Europe to meet enterprise data residency needs in regulated industries such as financial services, healthcare and government.

The partnership is notable because Elon Musk has previously criticized Anthropic, while SpaceXAI now provides compute capacity to one of xAI’s AI competitors.

Why This Matters

The deal shows how AI competition is moving beyond model releases into infrastructure control. For Claude users, the immediate result is higher usage capacity and fewer access constraints. For the wider AI market, the agreement underlines how compute availability, energy supply and chip access are becoming central factors in which companies can scale advanced AI products.

This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication.