SpaceX Integrates xAI, Rents 220K GPU Cluster to Anthropic

On May 6, Elon Musk announced that xAI will dissolve and merge entirely into SpaceX as a new AI product called SpaceXAI, according to Musk’s statement. Simultaneously, SpaceXAI announced an exclusive global compute partnership with Anthropic, in which the entire Colossus 1 supercluster will be leased to Anthropic for Claude series model inference services.

Integration and Prior Acquisition Context

The move marks the completion of AI business restructuring across Musk’s technology ecosystem. The integration follows SpaceX’s full acquisition of xAI in February 2025. According to reports, the transaction valued SpaceX at $1 trillion and xAI at $250 billion, with the combined entity reaching $1.25 trillion in valuation—reportedly setting a record for global high-valuation enterprise mergers. After the acquisition, xAI had maintained brief independent operations until this integration announcement.

Colossus 1 Specifications and Anthropic Partnership

Colossus 1, located in Memphis, Tennessee, is among the world’s largest and fastest-deployed artificial intelligence supercomputer clusters. The cluster is equipped with over 220,000 Nvidia GPUs, including high-density deployments of H100, H200, and next-generation GB200 accelerators, providing computational support for large language models, multimodal systems, scientific simulations, and generative AI at frontier scale.

Anthropicexpressed plans to leverage the added compute capacity to directly increase service capacity for Claude Pro and Claude Max subscription users. As part of the partnership agreement, Anthropic also expressed intent to collaboratively develop multi-gigawatt-scale orbital AI compute infrastructure.

SpaceX’s Orbital AI Compute Strategy

SpaceX stated that “the computational requirements for training and operating next-generation AI systems are growing at a rate far exceeding what ground-based electrical power, land, and cooling infrastructure can sustain, and the pace of expansion in these facilities cannot match critical timelines for AI development. SpaceX is the only institution with the requisite launch frequency, orbital delivery cost advantages, and constellation operations experience to transform orbital compute from a research concept into an engineering project deployable in the near term. If engineering challenges are overcome, orbital compute will provide near-infinite sustainable power while significantly reducing environmental impact on Earth.”

According to external reports citing IPO registration filings, SpaceX anticipates that over 90% of market opportunity may derive from the artificial intelligence sector.

Musk’s Assessment of Anthropic

Musk provided direct commentary on the partnership, stating: “Last week I spent considerable time with core members of the Anthropic team learning about their work to ensure Claude is beneficial to humanity, and I was impressed. Everyone I met is highly capable and genuinely cares about doing the right thing. No one triggered my ‘evil detector.’ As long as they maintain critical self-examination, Claude will likely be beneficial.”

Musk also disclosed that while Colossus 1 is leased to Anthropic, SpaceXAI’s training workload has migrated to Colossus 2.

Colossus 2 and Future Capacity

Colossus 2 is SpaceX’s globally first gigawatt (GW)-scale AI training supercomputer cluster, formally operational as of January 2026. The cluster will be fully equipped with next-generation GB200 GPUs and is projected to reach several times the scale of Colossus 1.

Background on Anthropic

Anthropicwas founded in 2021 by former OpenAI employees and develops the Claude series of large language models. Since its founding, the company has received investment from technology giants including Amazon, Google, Nvidia, and Salesforce. As of February 2025, the company’s valuation reached $380 billion. Anthropic’s Claude and xAI’s Grok are direct competitors in the general-purpose large language model market.