Google has signed a massive compute agreement with SpaceX. Under regulatory filings disclosed on Friday, Google will pay SpaceX $920 million per month from October 2026 through June 2029 for access to approximately 110,000 Nvidia GPUs, CPUs, memory, and related components. The total value of this contract is estimated at nearly $30 billion over its lifetime.
This milestone agreement lands just a week before SpaceX's historic IPO on Nasdaq, where it aims to raise $75 billion at a valuation of approximately $1.75 trillion. This is SpaceX’s second major compute deal in a month. In late May, Anthropic signed a deal worth $1.25 billion per month for the complete output of the Colossus 1 data center in Memphis, which was originally built by xAI for Grok before its merger into SpaceX.
Google's deal covers about half of the compute capacity that Anthropic secured. Both deals feature 90-day cancellation clauses after December 31, 2026. If SpaceX fails to deliver the promised GPU capacity by September 30, 2026, Google reserves the right to terminate the contract immediately after a one-month grace period, or accept reduced hardware at a proportionally lower rate.
The deal is highly notable as Google is traditionally considered a compute leader, possessing massive capacity through its custom TPU chips. A Google representative framed this SpaceX deal as a short-term bridge, stating it ensures "bridge capacity to meet surging customer demand for our agent platform, Gemini Enterprise, which has been even higher than we expected."
[AgentUpdate Depth Analysis] Google's decision to lease massive compute from SpaceX underscores a critical bottleneck in the AI Agent ecosystem: the explosive adoption of enterprise-level agents is outpacing the hyper-scalers' infrastructure roadmaps. Unlike simple chatbots, AI Agents executing long-horizon planning, tool use, and multi-turn reasoning require exponentially higher inference compute. Google's rely-on-TPU strategy, while cost-effective long-term, fell short in matching the immediate, surging demand for Gemini Enterprise. By tapping into SpaceX's Nvidia-rich Colossus infrastructure, Google highlights a shift toward heterogeneous, multi-cloud compute strategies for agent deployment. For the broader AI Agent landscape, this deal proves that the capability to orchestrate massive, low-latency compute pools dynamically is now the ultimate differentiator, positioning physical compute aggregation as a high-margin utility business.
