Anthropic and SpaceX Sign Compute Agreement Worth Up to $84.5 Billion, Terminable with 90 Days' Notice

AI giant Anthropic and SpaceX have signed a compute cooperation agreement worth up to $84.5 billion. The agreement includes a special clause allowing Anthropic to terminate the partnership with 90 days' notice, providing immense flexibility for AI companies' compute deployment.

With the exponential growth in compute demand for large model training and inference, securing stable GPU resources has become a core competitive advantage for AI companies. Meanwhile, Elon Musk is aggressively expanding his own data centers, aiming to bring over 1 million GPUs online by the end of the year, highlighting the massive investments and fierce competition among tech giants in compute infrastructure.

This collaboration not only provides potential compute guarantees for Anthropic's future large model R&D but also reflects the complex commercial ties between compute suppliers and demanders. The flexible termination clause implies that dynamic bargaining may exist in the AI compute market, requiring companies to constantly adjust their compute procurement strategies based on technological iterations and cost changes.

As relevant companies advance their IPO processes, the scale of investment in compute infrastructure and cooperation details are gradually becoming transparent. Such ultra-large-scale compute agreements indicate that the barriers to entry in the AI industry are continuously rising, and future compute reserves will directly determine the market landscape for large model developers.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.