Anthropic's $15B Annual Deal for Musk's AI Data Centers

Anthropic commits to paying $1.25 billion monthly through 2029 for access to SpaceX's Colossus data centers, revealed in SpaceX's IPO filing.
In a groundbreaking development within the artificial intelligence industry, Anthropic has entered into a substantial financial partnership with SpaceX to secure computing resources for its AI training operations. Earlier this month, both companies publicly announced their compute partnership, which grants the AI firm access to SpaceX's cutting-edge Colossus data centers located in Memphis, Tennessee. However, the true scope of this ambitious collaboration has only recently come to light with the detailed financial disclosures contained in SpaceX's official IPO filing, revealing the staggering scale of investment required to power next-generation artificial intelligence systems.
According to SpaceX's S-1 filing submitted to regulatory authorities, Anthropic agreed to pay $1.25 billion per month through May 2029 for exclusive access to SpaceX's AI training infrastructure at both Colossus I and Colossus II facilities. This commitment translates to a stunning $15 billion annual expenditure, underscoring the immense computational demands of training large language models and advanced AI systems at scale. To put this figure in perspective, this annual payment represents nearly double the $18.7 billion in total revenue that SpaceX reported across all its business operations throughout the entirety of 2025, highlighting just how critical and expensive high-performance computing infrastructure has become in the competitive AI landscape.
The magnitude of this financial commitment reflects the broader trend across the technology industry where major AI companies are increasingly willing to invest substantial capital to secure reliable, high-capacity computing resources. The data center partnership represents a critical bet by Anthropic that SpaceX's infrastructure can meet its computational needs for developing and improving its AI models. With GPU-intensive workloads and the exponential growth requirements of machine learning systems, securing dedicated access to purpose-built facilities has become a strategic imperative for companies advancing frontier AI technologies.
Source: The Verge


