Google to Pay SpaceX $920M/Month for 110K NVIDIA GPUs

On Friday, SpaceX announced in a regulatory filing that Google will pay $920 million per month from October 2026 through June 2029 for access to approximately 110,000 NVIDIA GPUs, CPUs, memory, and other components. The deal is similar in structure to SpaceX's recent agreement with Anthropic, which pays $1.25 billion per month for compute from the Colossus 1 data center near Memphis, Tennessee.

Google's payment is for roughly half the compute capacity that Anthropic secured. SpaceX did not specify which data center Google will use. CEO Elon Musk previously indicated that the Colossus 2 facility would be reserved for xAI (now part of SpaceX).

Why Google Needs the Compute

Google cited "unexpected demand" for its recently launched AI products. In a statement, a Google representative said: "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."

Despite being the world's largest single owner of AI compute, according to some estimates, Google is facing a capacity crunch. Parent company Alphabet has already committed to over $180 billion in capital expenditures this year and expects that to "significantly increase" in 2027. To fund this, Alphabet recently announced an $80 billion equity sale.

Cancellation and Ramp-Up Details

Both parties can terminate the agreement with 90 days' notice after December 31, 2026. Google's access to the data center will ramp up "through September at a reduced fee." If SpaceX fails to deliver the committed GPU count by September 30, 2026, Google may immediately terminate after a one-month grace period, or accept the reduced number with lower monthly fees.

Context: SpaceX's IPO and Data Center Ambitions

The deal was announced one week before SpaceX's stock is expected to begin trading on the Nasdaq. The company aims to raise about $75 billion at a valuation of ~$1.75 trillion, which would make it the largest IPO in history.

Google is a longtime investor in SpaceX. Its stake is expected to be worth over $100 billion after the IPO. The companies are also in talks to build orbital data centers, a key part of SpaceX's post-IPO plans.

Technical Implications

For developers, this deal signals that even hyperscalers like Google are struggling to keep up with AI compute demand. The reliance on external GPU clusters from a company like SpaceX—primarily known for rockets—highlights how critical hardware availability has become. If you're building on Google Cloud, expect increased capacity for GPU-intensive workloads like Gemini Enterprise, but also potential pricing adjustments as Google passes on these costs.

What This Means for Developers

  • Gemini Enterprise demand is real. Google is spending nearly $1B/month to meet it. If you're using Gemini APIs, expect continued investment in performance and reliability.
  • GPU scarcity persists. Even Google needs to rent from SpaceX. Plan your AI workloads accordingly—consider reserving capacity early.
  • Watch for pricing changes. Google's massive capex and this deal will likely lead to higher cloud GPU prices or tighter quotas.
  • SpaceX becomes a compute provider. Post-IPO, SpaceX's data center business could rival cloud providers. Keep an eye on their offerings.