The number sounds almost fictional: $920 million per month. That is what Google has agreed to pay SpaceX for computing power — not for a product, not for a service, but for raw machine infrastructure: GPUs, CPUs, and the physical hardware to run them. The deal, disclosed in a regulatory filing on June 5, 2026, is one of the most extraordinary commercial agreements in the history of the technology industry. And it tells you everything you need to know about where the AI race stands right now.
You do not need to be an investor or a cloud architect to understand the implications. If the world's most advanced AI companies are paying billions of dollars a month just to keep their systems running, the businesses that figure out how to leverage these tools strategically — even at a fraction of that scale — are the ones that will define the next decade.
The Deal in Plain Terms
Under the terms of the agreement, Google will pay SpaceX $920 million every month from October 2026 through June 2029 for access to approximately 110,000 Nvidia GPUs plus supporting CPUs and infrastructure. Google's parent company Alphabet has already committed more than $180 billion in capital expenditures in 2026 alone — and has signalled this figure will "significantly increase" in 2027. Even for a company of Google's scale, renting compute at this price reflects a genuine supply crunch: it does not have enough of its own infrastructure to meet current demand for AI services.
Why Google Went to SpaceX
This is the part that makes the deal genuinely surprising. Google is one of the world's largest single owners of AI compute — home to custom TPU chips, vast global data centres, and decades of infrastructure investment. Yet a Google spokesperson stated plainly that the SpaceX deal is "a short-term, timely agreement to ensure we have bridge capacity to meet surging customer demand" for Gemini Enterprise, its recently launched AI agent platform for businesses. That means even Google was caught off guard by how quickly demand for its AI products exploded. SpaceX, through its Colossus data centre originally built for Elon Musk's xAI models, is now monetising that infrastructure through compute-rental agreements at an extraordinary pace.
Anthropic Is Already in This Deal Too
Google is not the only tenant. Just two weeks before Google's announcement, Anthropic — maker of Claude — agreed to pay SpaceX $1.25 billion per month for full access to the Colossus 1 data centre, also running through 2029. Google's deal appears to cover roughly half the same facility's compute capacity. In practical terms, two of the biggest names in AI are now co-tenants in a data centre owned by Elon Musk, each paying sums that would fund entire national technology development programmes. The AI compute market has become its own industry — and SpaceX is rapidly becoming one of its most powerful landlords.
What $11 Billion a Year Tells You About AI's Trajectory
Run the numbers: $920 million per month is approximately $11 billion per year. That is just one company's rental bill for one set of data centres. It reflects a simple but staggering reality: AI training and inference at scale requires computational resources that the world has not finished building yet. The gap between AI ambition and available infrastructure is what is making compute the most contested commodity in the global economy right now — more strategically critical, in many respects, than oil. Every company building AI products is essentially competing for a finite pool of GPU capacity, and that competition is driving deals nobody would have believed five years ago.
What This Means for Nigerian and African Businesses
Here is the ground-level implication for businesses in Lagos, Abuja, Accra, and Nairobi. The AI tools you are already using — or considering adopting — are built on infrastructure wars happening at a scale you may never have to engage with directly. But you will feel the downstream effects:
- API pricing: As infrastructure costs fluctuate, AI API prices for tools like Gemini, Claude, and ChatGPT will shift — monitoring for pricing changes matters
- Enterprise AI adoption windows: Companies investing in AI agent platforms now, while early pricing is still competitive, are building structural advantages
- Data sovereignty: The concentration of compute in US data centres creates compliance conversations for African multinationals operating under NDPA and other local regulations
The Cancellation Clause Worth Watching
There is an important detail in both the Google and Anthropic deals: either party can terminate with 90 days' notice after December 31, 2026. That makes early 2027 a key test. If both companies renew, it validates SpaceX as a permanent player in global AI infrastructure. If they exit early, it signals that the capacity crunch is resolving faster than expected — and that the economics may be shifting. Either outcome tells you something important about the pace and direction of this market.
For now, the deal is the clearest possible signal of how seriously the world's largest technology companies are betting on AI. The question for your business: are you moving with the same urgency — even at a fraction of the scale?
Originally featured on TechCrunch