Google has agreed to pay SpaceX $920 million every single month for access to computing capacity inside xAI's data centers in Memphis, Tennessee. The contract runs from October 2026 to June 2029, adding up to roughly $30 billion in total. This is not a small or peripheral deal. It is one of the largest technology contracts signed in recent history.
The reason this is surprising is simple: Google is one of the largest builders of data centers on the planet. It has been spending tens of billions of dollars a year on its own infrastructure. So when a company at that scale pays a competitor's parent company nearly a billion dollars a month for borrowed computing capacity, you know something structural is going on.
What is going on is that demand for AI services is outpacing supply by a significant margin. Google Cloud described the deal internally as a way to build "bridge capacity" for its Gemini Enterprise product, the AI subscription it sells to large businesses. Demand grew faster than Google could build. So it rented.
SpaceX, which merged with Musk's AI company xAI in February this year, now has two of the world's biggest AI companies paying it rent. Anthropic, which makes the Claude AI assistant and competes directly with Google, is paying SpaceX $1.25 billion a month for access to the Colossus 1 data center, the facility that xAI originally built to train its own Grok AI models. Now that xAI has moved its own training work to the newer Colossus 2 facility, Colossus 1 is being leased out to Anthropic. The 110,000 chips Google is accessing appear to be part of the expanded infrastructure.
The Colossus facility in Memphis is itself a remarkable story. xAI built the original supercomputer in 122 days by converting a former Electrolux appliance factory, a project that would normally take two years or more. The site has expanded rapidly and now houses hundreds of thousands of chips across multiple buildings.
There is a financial irony in all of this. xAI posted a $6.4 billion operating loss last year on revenue of $3.2 billion. It is burning through cash at a rate that concerned some investors ahead of the SpaceX IPO. Signing Google and Anthropic as paying tenants transforms that liability into a selling point: xAI's infrastructure is now a revenue source, not just a cost center.
This is the core reason these deals were disclosed in SpaceX's IPO filings. SpaceX plans to list on Nasdaq on June 12, targeting a $1.77 trillion valuation and aiming to raise $75 billion, which would make it the largest IPO in stock market history. The Google and Anthropic contracts are central to the investment case: they show that the infrastructure SpaceX built for its own AI ambitions can also function as a commercial business.
For business operators outside the AI industry, the practical signal here is about cost and availability. The computing resources required to run large AI products are genuinely scarce right now. Roughly half of planned new data centers in the US for 2026 are expected to be delayed or cancelled, held back by shortages of power grid equipment, not chips. The companies that control existing, operational capacity have significant leverage.
If your business is evaluating AI tools or platforms, the companies behind those tools are in an infrastructure arms race where access to computing power shapes what they can offer and at what price. Google paying $920 million a month to keep its AI products running smoothly is a reminder that the costs underlying these services are enormous, and those costs will eventually flow through to enterprise pricing.
The relationship between Google and SpaceX is also more intertwined than the headlines suggest. Google held roughly a 6% stake in SpaceX as of end 2025, before the xAI merger diluted that to around 5%. The two companies are also reportedly discussing SpaceX's potential role in launching hardware for Google's orbital data center experiments. Competitors, landlord and tenant, and partial co-owners, all at the same time.