Google signed a cloud services agreement with SpaceX on June 5 to rent access to roughly 110,000 NVIDIA GPUs for approximately $920 million per month. The contract runs from October 2026 through June 2029 and carries an annual run-rate of about $11 billion, comparable to Anthropic’s total annual compute budget. Winbuzzer reported the terms on June 6.
The framing Google chose is revealing. The company described the arrangement as “bridge capacity” to meet stronger-than-expected demand for Gemini Enterprise, its agent platform for business customers. Google has more in-house AI infrastructure than any other AI vendor. Paying $11 billion a year to rent from a third party is not a routine procurement decision; it is a public admission that demand is outrunning the build schedule.
Alphabet’s own capital plans make the context sharper. Google expects $180 billion to $190 billion in capital expenditures for 2026, overwhelmingly for technical infrastructure, and has proposed an $80 billion equity raise to expand compute further. The SpaceX arrangement sits alongside that buildout, not instead of it. Google needs the GPU block for roughly three years while its longer-term infrastructure programs catch up to Gemini Enterprise demand.
The contract structure includes real protections for Google. SpaceX must deliver the committed GPU access by September 30, 2026. If it misses that deadline, Google can terminate after a one-month grace period or accept reduced capacity at a proportionally lower fee. After December 31, 2026, either party can exit with 90 days’ notice. The monthly fee is tied to delivered hardware, not promised hardware. That clause matters because SpaceX’s supply chain for 110,000 GPUs at this timeline is not trivial.
For SpaceX, the contract is significant for a different reason. SpaceX acquired xAI in February 2026 and, with it, the Colossus 1 data center. Anthropic agreed in May to rent compute from that same facility. Google’s agreement adds a second major outside customer to SpaceX’s compute estate and strengthens the recurring-revenue case SpaceX needs ahead of a rumored IPO targeting a valuation above $1.75 trillion, a figure that outside analysts have disputed.
The structural shift worth tracking is that SpaceX is now, credibly, a third compute infrastructure provider alongside AWS and Azure. Two anchor customers at nine-figure monthly contracts gives SpaceX a negotiating position that no other challenger to the hyperscalers has reached. For Anthropic, OpenAI, and any other lab that sources frontier compute externally, SpaceX entering the market at scale changes the supply side of a conversation that has been dominated by two vendors for years.
Google Cloud’s backlog has grown past $460 billion, with just over half expected to be recognized within the next 24 months. That backlog number helps explain the urgency behind the SpaceX deal: the revenue commitment is already on the books, and Google needs the compute to service it.
Any lab currently negotiating multi-year compute contracts with AWS or Azure should factor SpaceX’s pricing signal into the conversation before the next renewal date.
Source: Winbuzzer (winbuzzer.com), published June 6, 2026.
