Anthropic Secures Colossus Data Center Capacity from SpaceX/xAI, Raising Environmental and Supply Chain Concerns

Summary

Anthropic announced a major infrastructure deal at its Code w/ Claude event, securing all of the capacity of xAI/SpaceX's Colossus 1 data center. The partnership addresses Anthropic's critical compute constraints while allowing xAI to continue training its own models on the larger Colossus 2 facility. However, the move has drawn criticism due to the Colossus data center's documented environmental record and introduces potential supply chain vulnerabilities.

The Colossus data center has faced scrutiny for operating gas turbines without Clean Air Act permits or pollution control devices, by classifying them as "temporary" installations. Reports have linked the facility to increased hospital admissions related to poor air quality. Prominent voices in the data center discussion, including Andy Masley, have explicitly cautioned against using this facility, given the escalating political sensitivity around AI data centers.

Beyond environmental concerns, the deal introduces a unique supply chain risk: Elon Musk has effectively given himself the power to reclaim Anthropic's compute access if he determines the company's AI "engages in actions that harm humanity." While framed as alignment-focused stewardship, this arrangement concentrates significant control over critical infrastructure in a single individual's hands, creating both operational and strategic vulnerability for Anthropic.

• xAI retains larger Colossus 2 facility for its own training, contradicting early speculation that the deal meant abandonment of Grok models