Anthropic, the AI research company behind the Claude family of models, is paying SpaceX $1.25 billion per month for access to computing infrastructure. The deal runs through May 2029.

Inside the deal

The arrangement gives Anthropic full access to SpaceX’s Colossus 1 data center, which houses more than 220,000 NVIDIA GPUs and delivers over 300 megawatts of AI compute power. Anthropic is using it to train and serve language models.

The Colossus 1 facility was originally built to support xAI’s Grok models. SpaceX is now monetizing what appears to be underutilized infrastructure, turning a cost center into a revenue stream that could exceed $15 billion annually from a single customer.

Anthropic has doubled five-hour rate limits across its Pro, Max, Team, and Enterprise plans. Peak-hour usage caps have been removed entirely. API rate limits for Claude Opus models have been significantly increased.

Why an AI company is buying from a rocket company

The two companies are also reportedly exploring the development of multiple gigawatts of orbital AI compute capacity.

What this means for crypto and AI investors

Decentralized compute networks like Render, Akash, and io.net have positioned themselves as alternatives to centralized GPU providers. A $15 billion annual deal from a single AI customer underscores the scale of demand these protocols are trying to capture.

The Colossus 1 data center runs on NVIDIA GPUs. Every major AI compute agreement further entrenches NVIDIA’s hardware as the industry standard, which has downstream effects on crypto mining hardware markets and GPU-dependent proof-of-work chains.

The deal also raises questions about concentration risk. Anthropic is betting its compute future on a single facility controlled by a SpaceX entity that also supports a competing AI product in Grok. If that relationship sours, or if xAI reclaims capacity, Anthropic could find itself back in the same compute-constrained position that drove this deal in the first place.