Google is moving to rent massive AI computing capacity from SpaceX, a sign that the race to power artificial intelligence has grown so intense that even the biggest tech companies are struggling to keep up.
According to a SpaceX regulatory filing, Google has agreed to pay US$920 million per month from October 2026 through June 2029 to access a huge package of computing resources, including about 1,10,000 NVIDIA GPUs, CPUs, memory, and related components. The deal does not give Google ownership of the hardware, but it does secure access to the computing power needed to run its AI services at scale.
Why Google Is Renting Instead of Building
The arrangement highlights a timing problem that is now shaping the AI industry. Google can spend heavily, but building new data centers takes years, while demand for AI products is rising right now.
Google has already committed to more than US$180 billion in capital spending this year, and it said spending will rise significantly in 2027. The company also recently announced a US$80 billion equity sale to help fund its expansion plans.
Even with that level of investment, Google told TechCrunch that the agreement will help meet surging customer demand for its AI offerings, including agent platforms and Gemini Enterprise.
The Real Bottleneck Is Infrastructure
The SpaceX deal shows that the most valuable asset in AI is no longer just the model itself. The infrastructure required to train and run those models has become a major competitive battlefield.
Advanced AI systems need enormous numbers of specialized chips, especially NVIDIA GPUs, and demand for those chips has climbed as AI adoption spreads. That pressure is pushing more companies toward rental and partnership models because they can be faster than waiting for new facilities to come online.
Anthropic has already taken a similar path, signing an agreement to use SpaceX’s Colossus 1 supercomputer for its own AI workloads.
How SpaceX Turned Spare Capacity Into Revenue
The business opportunity at SpaceX emerged after the company merged with Elon Musk’s AI firm, xAI, gaining control of large-scale AI infrastructure, including the Colossus data center cluster originally built to support the Grok chatbot and other AI projects.
Instead of letting unused computing capacity sit idle, SpaceX has chosen to lease out the extra capacity. The company says the approach gives it flexibility to allocate and monetize the infrastructure while generating direct revenue from hardware that xAI has not yet fully used.
The strategy appears to be paying off quickly. Google’s deal comes only months after Anthropic agreed to pay US$1.25 billion per month through 2029 for access to computing capacity at Colossus 1 near Memphis, Tennessee.
Together, the two contracts push SpaceX’s AI-related infrastructure revenue above US$2 billion per month.
What the Contract Means for Google and SpaceX
The agreement also matters for SpaceX as it prepares for what could become one of the largest IPOs in history. The company is reportedly targeting a valuation of more than US$1.75 trillion.
Long-term infrastructure contracts with recurring payments could make SpaceX more attractive to investors by providing a steadier and more predictable revenue profile ahead of a public listing.
The deal includes protections for Google as well. SpaceX must deliver the promised computing capacity by 30 September 2026.
If that does not happen on time, Google can either terminate the agreement after a one-month grace period or accept fewer GPUs at a correspondingly lower cost.
The contract also allows either side to cancel the deal with 90 days’ notice after 31 December 2026. In practice, the arrangement makes clear that access to computing power has become just as strategically important as building the AI models themselves.