Anthropic Agrees to Pay SpaceX Up to $84.5 Billion for Compute Through 2029

Anthropic has agreed to spend up to $84.5 billion on AI computing resources from SpaceX through 2029, as the company expands the infrastructure needed to train and operate advanced artificial intelligence models.

Ad

Ad

Anthropic Signs Large SpaceX Compute Agreement

Anthropic revealed the deal in confidential IPO documents, according to the filing information reported in the documents. The agreement provides access to computing power on Nvidia chips for the AI firm.

The agreement could reach $84.5 billion in total spending through 2029. However, Anthropic can reportedly cancel most of the contract with 90 days’ notice.

That structure provides Anthropic with greater flexibility than a few of its other long-term infrastructure agreements. The firm has stated that having adequate computing power could be crucial for the future development of AI.

Anthropic expects demand for advanced AI systems to continue rising. As a result, the company is securing large amounts of capacity from several technology and infrastructure providers.

Meanwhile, Claude Opus 5.5 belongs to Anthropic’s new Claude 5.5 model family and supports coding, research, and professional knowledge work. Anthropic said the model delivers performance close to Claude Fable 5.1 while reducing costs for common workloads.

Ad

Ad

Compute Commitments Reach Hundreds of Billions

Anthropic expects to spend at least $518 billion on AI infrastructure over the coming years. The company has agreements involving Google, Amazon, Microsoft, Broadcom, AMD, and other partners.

Its long-term commitments include at least $111.1 billion with Google and $110 billion with Amazon. Another deal with Microsoft is worth approximately $31.4 billion for infrastructure services.

Anthropic also has about $161.2 billion in equipment leases tied up with Broadcom. Many of those agreements have limited cancellation options, and Anthropic must make payments under specific terms.

The company has also started moving beyond a cloud-only model. Its infrastructure plan has evolved to encompass exclusive data centers, outright chip leases, and long-term access to computing resources.

Anthropic Expands Infrastructure Ahead of IPO

Anthropic confidentially filed for a U.S. initial public offering in June. The company has not yet made the complete prospectus available to the public, which means the terms of its agreements are only available in part.

Additionally, Anthropic has reportedly delayed its IPO to November, with the offering potentially ranking among the largest ever if fundraising targets are met.

The AI startup reported nearly $4.6 billion in revenue for 2025. Meanwhile, operating losses surged to over $8 billion, while spending on infrastructure and model development was high.

Anthropic held about $20.28 billion in cash, cash equivalents, and short-term investments at the end of 2025.

The SpaceX agreement adds another large source of compute to Anthropic’s infrastructure plans. Final spending will depend on how much capacity Anthropic uses and whether it keeps the agreement through 2029.