Anthropic⁠ has reached an agreement to use computing infrastructure operated by SpaceX⁠, marking a significant partnership between the artificial intelligence company and Elon Musk despite his past criticism of the firm.

Massive AI Computing Expansion

Under the agreement announced Wednesday, Anthropic will gain access to the full computing capacity of SpaceX’s Colossus 1 facility in Memphis, which reportedly houses more than 220,000 Nvidia processors.
The deal is expected to provide Anthropic with an additional 300 megawatts of computing capacity within a month, as the company seeks to expand performance for its Claude AI services, including Claude Pro and Claude Max subscription plans.

Anthropic said the increased infrastructure would allow the company to remove peak-hour usage limits for paid users and significantly expand developer access to its Claude Opus models.

New ‘Dreaming’ Feature Unveiled

The announcement coincided with Anthropic’s developer conference, where the company introduced a research-preview feature called “dreaming.”

According to the company, the feature enables AI systems to review tasks between sessions, recognize patterns and update contextual memory files used to personalize responses.

Anthropic also introduced new tools for managing AI agents, software systems designed to perform tasks with limited human input.

Potential Orbital Data Centers

Anthropic said it is exploring cooperation with SpaceX on future space-based orbital data centers, an idea long promoted by Musk as part of broader plans to expand computing infrastructure beyond Earth.
Such projects would likely require major investment and significant technical development.

Musk Signals Support for Anthropic

In a post on X, Musk said he agreed to lease computing resources to Anthropic after meeting with the company’s leadership.
“Their work to ensure Claude AI is good for humanity impressed me,” Musk wrote, adding that “no one set off my evil detector.”

Broader AI Competition

The partnership comes amid intensifying competition among major AI firms including OpenAI⁠, Google⁠, Microsoft⁠ and Musk’s own AI venture xAI.

Musk remains involved in legal disputes with OpenAI and CEO Sam Altman over the company’s structure and direction.

Meanwhile, the U.S. government has recently expanded cooperation with several AI companies on cybersecurity and defense-related initiatives, though Anthropic has reportedly not participated in some of those agreements.