Anthropic has agreed to pay SpaceX $1.25 billion per month through May of 2029 for access to cloud computing infrastructure, a long-awaited US regulatory filing revealed on Wednesday. In other words, Anthropic will be sending a rival artificial intelligence lab roughly $15 billion a year, an extraordinary sum that demonstrates how access to compute has become one of the defining bottlenecks in the race to develop advanced artificial intelligence.
Anthropic and SpaceX announced a deal earlier this month that gives the Claude developer access to GPUs at Colossus and Colossus II, a pair of data centers straddling Tennessee and Mississippi with more than one gigawatt of computing power. SpaceX had rushed to build the facilities for its xAI unit, which develops the Grok AI chatbot, but Musk said his company didn’t need all of their computing capacity in the end. Terms of the deal had not been previously disclosed.
Anthropic is paying an unspecified reduced fee for May and June before the $1.25 billion per month rate takes effect, SpaceX said in its S-1 regulatory filing. The eye-popping figure is a sign of how hungry Anthropic is for computing resources needed to power products like its increasingly popular AI coding tools. The company’s revenue for the second quarter of 2026 is expected to exceed $10 billion, according to The Wall Street Journal.
An Anthropic spokesperson confirmed the figures to WIRED. SpaceX did not immediately respond to WIRED’s request for comment. In a post on X about the deal, Musk said that SpaceX was “in discussions with other companies to do the same.”
SpaceX says it expects to “enter into additional similar services contracts” for its compute infrastructure and will continue using its data centers for itself. “We have sufficient capacity to provide compute for our own AI models, including support of our training and inference demands, and to satisfy the obligations under these agreements,” the filing states. “We believe our dual monetization strategy provides multiple pathways to generate returns on invested capital.”
The filing details SpaceX’s business opportunities and risks ahead of an initial public offering. SpaceX is pursuing the largest IPO in history, with hopes of raising about $75 billion at a valuation of $1.75 trillion. The company filed its initial paperwork confidentially with the US Securities and Exchange Commission on April 1. The filing released on Wednesday is the cleaned-up version, though additional changes could come before it debuts on the Nasdaq stock exchange under the ticker SPCX, which could reportedly come as soon as June 12.
[AgentUpdate Depth Analysis] This monumental transaction underscores the intense capital intensity and severe compute bottlenecks defining the frontier AI landscape. Anthropic’s massive commitment to secure Colossus-class infrastructure illustrates that algorithms alone are no longer the primary differentiator; raw, high-throughput compute is the ultimate leverage. For the AI Agent ecosystem, this deal signals a looming shift in cost dynamics. As Agents transition from simple text generation to autonomous, long-horizon multi-agent workflows running continuous tasks, inference costs will skyrocket. Startups relying on API-driven agentic architectures will face immense margin pressure, while players controlling vertically integrated compute fabrics will hold an asymmetric advantage in scaling the next generation of autonomous digital workers.
