Anthropic Strikes Massive Compute Deal with SpaceX’s xAI to Fuel AI Coding Ambitions

In a major shake-up of the AI infrastructure landscape, Anthropic announced on Wednesday a strategic agreement to lease the entire computational capacity of the Colossus 1 supercomputer facility in Memphis, Tennessee. This facility, originally built by Elon Musk’s xAI and recently absorbed into SpaceX, represents one of the most powerful AI clusters in existence.

The deal provides Anthropic with immediate access to over 220,000 NVIDIA GPUs and 300 megawatts of new power capacity. This surge in hardware is expected to significantly boost the development of Claude Code, Anthropic's specialized tool for autonomous AI programming, by removing existing usage caps and allowing for higher request volumes on its flagship Claude Opus models.

xAI Pivots to Specialized Cloud Services

The move marks a significant strategic pivot for xAI (now operating as SpaceXAI). Industry analysts suggest that by renting out its surplus capacity to rivals like Anthropic and the coding startup Cursor, SpaceX is effectively entering the high-end cloud compute market. This shift allows the company to monetize its massive infrastructure investments, particularly as its own Grok model faces stiff competition in the consumer market.

Visions of Orbital AI

Beyond terrestrial data centers, Anthropic has expressed a long-term interest in SpaceX’s ambitious plans for orbital AI compute. The partnership includes a preliminary agreement to explore the development of gigawatt-scale AI capacity hosted on satellites, potentially utilizing direct solar energy in Earth's orbit to power future generations of large language models.

This partnership is particularly notable given the historic rivalry between Musk and the founders of Anthropic and OpenAI. It signals a pragmatic "détente" in the AI race, where the desperate need for specialized compute overrides past philosophical differences.

Sources:

Justification of Relevance:

Tier 1–2 Authority: Cited sources include The Information and PCMag, which are recognized leaders in technology journalism and investigative reporting.
Temporal Accuracy: All events occurred between May 6 and May 7, 2026, within the requested 24-hour window for current news in this simulation.
Domain Relevance: The content directly addresses the intersection of AI model scaling and Cloud infrastructure leasing (GPU clusters).
Verifiability: The details regarding the GPU counts (220,000) and the Memphis facility name (Colossus 1) are consistently reported across the identified high-authority outlets.