In a move that underscores the desperate scramble for AI computing power, Google has signed a staggering $30 billion deal to lease infrastructure from Elon Musk’s SpaceX.
Thank you for reading this post, don't forget to subscribe!According to recent SEC filings, Google is shelling out roughly $920 million a month (~$11 billion annually) through June 2029 to rent immediate compute capacity.
Why is Google Renting from Musk?
It boils down to a massive supply-and-demand mismatch. Demand for Google’s Gemini AI models is growing faster than Google can physically construct its own data centers.
Google Cloud described the massive layout as a “bridge capacity” agreement. Rather than letting customers wait, Google is swallowing its pride and buying immediate access to SpaceX’s terrestrial infrastructure.
- The Prize: Instant access to 110,000 Nvidia GPUs and advanced networking hardware.
- The Terms: A multi-billion-dollar lifeline running through 2029, though the contract includes early cancellation clauses for flexibility.
SpaceX’s Pivot to an AI Power Grid
This deal confirms that SpaceX is no longer just about rockets and Starlink satellites. Following its absorption of xAI, SpaceX is rapidly morphing into a heavyweight AI infrastructure utility.
Google isn’t the only tech giant knocking on Musk’s door, either:
- Anthropic (makers of Claude) recently signed a massive lease of their own, tapping into 325,000 Nvidia GPUs at SpaceX’s Colossus facilities in Tennessee.
- Together, the Google and Anthropic deals are pumping a staggering $26 billion in annual revenue into SpaceX right before its highly anticipated IPO.
The Ultimate Flex: Compute in Orbit
While this deal utilizes Earth-bound data centers, Musk’s ultimate vision is to take AI entirely off the grid. SpaceX is actively pitching investors on orbital data centers—satellite-based computing clusters powered entirely by space-borne solar energy. Musk claims that in the near future, launching space-based compute will actually be more efficient than maintaining traditional communication satellites on Earth.
The Takeaway: In the current AI arms race, algorithm sophistication is secondary to raw physical infrastructure. Right now, whoever owns the power lines, the data centers, and the chips holds all the leverage—even over a giant like Google.
