SpaceX to grant Anthropic access to its Colossus AI supercomputer

Elon Musk’s SpaceX has entered into a significant infrastructure agreement with AI firm Anthropic, granting Claude’s developer access to the full compute capacity of the Colossus 1 data center in Memphis, Tennessee, amid growing competition in AI infrastructure.

Anthropic confirmed this partnership, stating it will grant the company over 300 megawatts of additional compute power and access to more than 220,000 NVIDIA GPUs within a month.

The company emphasizes that this expansion aims to meet the rising demand for its Claude AI products, including Claude Code, an AI coding assistant. The Colossus 1 facility was initially built to support Musk’s AI ambitions through xAI, which was integrated into SpaceX earlier this year as part of a broader restructuring of Musk’s AI and space ventures.

Following the agreement, Anthropic also announced increased usage limits for Claude Pro and Claude Max subscribers, highlighting how access to large-scale compute infrastructure is increasingly influencing the AI industry. The partnership occurs as AI companies race to secure GPUs, data centers, and power capacity to support more advanced AI models and enterprise applications.

Industry attention is shifting toward compute availability to scale generative AI tools, coding assistants, and autonomous AI systems. Reuters also reported that Anthropic introduced a new “dreaming” feature for Claude at its developer event in San Francisco.

This feature enables AI agents to analyze work across sessions and adapt to user contexts over time.

Additionally, reports indicate that Anthropic and SpaceX are exploring long-term plans for orbital AI data centers, aligning with Musk’s broader goals for space-based computing infrastructure.

The financial terms of the agreement were not disclosed.