Google just signed a $30 billion multi-year cloud computing agreement with SpaceX not for rockets, not for satellites, but for raw GPU compute capacity. When one of the world’s most powerful technology companies needs to rent processing power from a rocket company, it tells you everything about how desperate the AI infrastructure scramble has become.
The numbers are staggering. At full scale, Google will pay approximately $920 million per month to SpaceX from October 2026 through June 2029. In exchange, Google gets access to approximately 110,000 Nvidia GPUs plus CPUs, memory, and supporting infrastructure compute capacity Google needs to keep up with exploding demand for Gemini Enterprise and its broader AI services while its own data center expansion catches up.
The deal includes a ramp-up period through September 2026 before full capacity kicks in. Google reportedly retains the right to terminate if SpaceX fails to deliver the promised GPU capacity on schedule, a sensible protection given the scale of the commitment.
For SpaceX, this is a strategic masterstroke entirely separate from its space business. Combined with an earlier computer agreement with Anthropic, SpaceX’s total AI infrastructure contract value now exceeds $70 billion, a compelling financial narrative for a company preparing for a major IPO.
The broader signal is the real story. AI competition has shifted from who builds the best models to who controls enough compute power to run them at scale. GPUs are the new oil and the companies that control them are writing billion-dollar contracts to prove it.
Discover more from Phoonomo
Subscribe to get the latest posts sent to your email.
