Google and SpaceX are in early talks to put AI data centers into orbit, according to TechCrunch AI, citing a Wall Street Journal report. The discussions land just as SpaceX preps for a reported $1.75 trillion IPO later this year, with orbital compute pitched to investors as the cheapest place to run AI workloads within a few years. This is significant because it pushes the AI infrastructure arms race off the planet entirely.
What’s on the table
The reported partnership would explore launching data centers into space, with SpaceX providing the lift and Google bringing the compute. TechCrunch AI reports Google is also shopping the idea around to other rocket-launch companies, not betting solely on SpaceX. Google isn’t new to this. Late last year it announced Project Suncatcher, an initiative to launch prototype satellites by 2027.
There’s also a recent precedent in the orbit: Anthropic struck a deal last week with SpaceX to tap compute from xAI’s Memphis data center, with future orbital collaboration on the table. SpaceX acquired xAI back in February, which makes that stack a lot more interesting.
Why orbital compute, why now
Elon Musk has been the loudest voice arguing that space-based data centers will be cheaper to run. The pitch:
- Solar power around the clock, no grid hookups required
- Passive cooling in the vacuum of space
- No zoning fights, no community pushback, no local water disputes
That last point matters more than it sounds. Ground-based data center buildouts in the U.S. are running into serious local resistance over power draw, water use, and noise. Orbit sidesteps all of it.
The economics don’t quite work yet
Here’s the catch. TechCrunch AI recently reported that today’s terrestrial data centers remain significantly cheaper than orbital ones once you factor in satellite construction and launch costs. The math only flips if launch prices keep falling, which is exactly what SpaceX is selling investors on with Starship.
Google has skin in that game already. The company invested $900 million in SpaceX back in 2015, per regulatory filings. So a Google-SpaceX orbital partnership isn’t a cold start. It’s a decade-old relationship moving into a new phase.
What to watch
A few things stand out here for anyone tracking AI infrastructure:
- The IPO narrative. SpaceX needs a story big enough to justify a $1.75 trillion valuation. Orbital AI compute is that story.
- Project Suncatcher’s 2027 timeline. Google’s prototype satellites will be the first real-world test of whether the unit economics can work.
- The xAI-Anthropic-SpaceX triangle. Compute alliances are getting weird. Expect more of them.
Google and SpaceX haven’t commented publicly on the talks. For now, treat this as a signal of where hyperscaler thinking is headed, not a done deal. More details at the original TechCrunch AI report.