Why it matters
The strategic read
Adds Google to the small set of hyperscalers buying capacity from SpaceX/xAI rather than only building internally. The deal turns Colossus-era GPU capacity into a third-party cloud-service business line, showing even Google is willing to rent a six-figure NVIDIA GPU block when time-to-compute matters.
Source-supported terms
What was disclosed
- 01
SpaceX filed a June 5, 2026 free-writing prospectus disclosing a Cloud Service Agreement with Google LLC.
- 02
The compute capacity includes approximately 110,000 NVIDIA GPUs, CPUs, memory and other related components.
- 03
Google agreed to pay SpaceX $920M per month from October 2026 through June 2029.
- 04
Capacity ramps through September 2026 at a reduced fee.
- 05
If SpaceX fails to deliver access to the committed GPU amount by September 30, 2026, Google may terminate after a one-month grace period or accept fewer GPUs with a pro rata fee reduction.
- 06
After December 31, 2026, either party may terminate the agreement on 90 days notice.
- 07
Google retains ownership and intellectual-property rights in its content, AI models and related data.
- 08
TechCrunch reported the agreement as a new Google compute deal that is similar in length and scope to SpaceX's already-disclosed Anthropic Colossus compute agreements.
Disclosure boundary
What is known—and what is not
Primary SEC free-writing prospectus summary plus TechCrunch market reporting. Public details disclose parties, term, monthly fee, GPU count, ramp deadline, delivery-failure remedy, termination right and IP/data ownership. The full cloud-service agreement, exact data-center allocation, GPU generation mix, networking topology, SLA terms, power/cooling allocation, workload purpose, pricing offsets, security terms and capacity-by-phase schedule are not public.