Google’s $920M SpaceX compute deal: from rockets to compute-as-a-service
Google signed a deal worth $920 million per month for compute capacity from SpaceX, covering 110,000 Nvidia GPUs plus supporting hardware, running from October 2026 through June 2029.
As a sophisticated cloud buyer, this move from Google puts eyes on what's potentially the next major cloud provider and the bigger picture being where future investments may flow as development environments take a major shift.
There's a growing general consensus that there's a storage shortage that even hyperscalers and major AI labs have to source storage from third parties.
Google already has its own TPU infrastructure and uses AWS, Azure, and GCP at massive scale.
Understanding fresh demand
Estimates believe Google to be the world's largest single owner of AI compute, yet it is faced with what it calls “short-term constraints.”
In a statement, Google described the deal as a result of unexpected demand for its recently launched AI products: "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.”
Alphabet CEO Sundar Pichai was explicit: "We are compute constrained in the near term. Our cloud revenue would have been higher if we were able to meet the demand.”
This tells us that this is an AI demand problem and the reason for this isn't necessarily because users need to generate images of cats or videos of dogs but because AI itself is becoming a fundamental infrastructure layer, soon to be central to major and all businesses.
This new reality creates demand from several industries. As such, businesses like Google, who are major cloud players serving several businesses, will be faced with demand that cannot be filled by building new data centers as that takes years, a costly option to renting storage from other providers.
A rocket company turned cloud provider
What are the reasons SpaceX is today in the news as a cloud provider?
The infrastructure origins trace back to xAI, Elon Musk's artificial intelligence venture whose gen AI, Grok, is largely integrated into the X social platform we all know previously as Twitter.
xAI built the Colossus 1 facility to train and run its Grok models. When SpaceX and xAI merged in February 2026, the compute assets came along with it.
What followed this merger was a deliberate strategic shift where instead of keeping GPU capacity dedicated to internal workloads alone, SpaceX moved toward leasing it externally, effectively becoming an AI cloud provider alongside Amazon Web Services, Microsoft Azure, and Google Cloud.
SpaceX has also recently been in the news for a compute deal with Anthropic, the most valuable private AI company following recent fund raising.
The combined annual value of the Google deal and the earlier Anthropic deal alone exceeds $25 billion, surpassing SpaceX's entire 2025 revenue from Starlink, launch services, and AI combined.
A rocket and satellite company is now on track to be primarily a compute-as-a-service business by revenue, this tells us everything we need to know about what the future will look like.
Posted Using INLEO