Even the world's largest AI compute operator is renting capacity from a rival's infrastructure, signaling that GPU demand has outpaced even Alphabet's $180B capex plan. SpaceX becomes the compute landlord to both Google and Anthropic weeks before its IPO.
Key Takeaways
- $920M per month is what Google will pay SpaceX for roughly 110,000 NVIDIA GPUs from October 2026 through June 2029
- Anthropic pays $1.25B/month for the same period under a similar deal, giving SpaceX two of the biggest compute contracts in history
- Google calls it bridge capacity for Gemini Enterprise, whose demand has exceeded internal forecasts
- A cancellation clause lets either party exit with 90 days’ notice after December 31, 2026
What Happened
Google has agreed to pay SpaceX $920 million per month for compute access, disclosed via a regulatory filing published June 5. The contract runs from October 2026 through June 2029.
Under the terms, Google gets access to approximately 110,000 NVIDIA GPUs, CPUs, memory, and related components. SpaceX did not name the specific data center involved, though CEO Elon Musk has previously indicated that Colossus 2 is reserved for xAI.
The deal closely mirrors the one Anthropic struck with SpaceX in late May, paying $1.25 billion per month for the full compute capacity of the Colossus 1 data center near Memphis, Tennessee. Google’s arrangement appears to cover roughly half that volume.
Google’s access will ramp up through September at a reduced fee. If SpaceX fails to deliver the committed GPU count by September 30, 2026, Google may terminate or accept a reduced allocation at lower monthly fees following a one-month grace period.
Why It Matters
Alphabet has already committed more than $180 billion in capital expenditures this year and is running an $80 billion equity sale to fund further infrastructure. That it is still renting GPU capacity from a competitor’s data center points to a genuine, acute gap between AI demand and available supply, even for a company of its scale.
The skeptic’s read is that this is a bridging move with a soft exit built in, not a structural shift. Google’s own statement frames it as short-term capacity for Gemini Enterprise rather than a strategic dependency on SpaceX. With cancellation available after December 31, 2026, the deal could last as little as three months on Google’s side if internal capacity catches up faster than expected.
This is a short-term, timely agreement to ensure we have bridge capacity to meet surging customer demand for our agent platform, Gemini Enterprise, which has been even higher than we expected. — Google spokesperson, Alphabet
Bottom Line
Watch whether SpaceX’s compute landlord role expands further after its Nasdaq listing, which is expected within a week. The company is targeting a $75 billion raise at a roughly $1.75 trillion valuation, and deals like this one position compute leasing as a recurring revenue line alongside rockets.
For teams evaluating AI infrastructure strategy: if Google is renting capacity at $920 million a month because internal build timelines cannot keep up, the enterprise lesson is that GPU access is now a procurement variable, not a given. Vendor lock-in risk and contract flexibility clauses deserve the same scrutiny as SaaS pricing.
