Google's AI Chips Head to Orbit Aboard SpaceX Falcon 9

Alphabet plans to launch its homegrown AI chips into orbit on Thursday aboard a SpaceX Falcon 9 rocket, marking the first in-orbit test of Project Suncatcher — the company's moonshot bid to run AI computing infrastructure in space.

The launch is scheduled for 11:15 am PT from Vandenberg Air Force Base in California's Santa Barbara County. During the uncrewed Transporter-18 mission, SpaceX will carry Planet Labs satellites, including a solar-powered prototype equipped with Google's tensor processing units, or TPUs.

The stakes for the AI industry are considerable. Data center construction on the ground has run into mounting opposition over power consumption, land use, and grid strain. SpaceX has promoted supercomputers in space as the next frontier in AI infrastructure, touting what it calls "infinite real estate" in orbit as a means of escaping the data center backlash on Earth. Alphabet's launch is the first concrete hardware test of whether that pitch can survive contact with orbital reality.

Project Suncatcher

Google first revealed Project Suncatcher in November 2025. The initiative's goal is to develop reliable, solar-powered AI computing infrastructure that can operate continuously in space. In a post announcing the launch plan, Alphabet described Project Suncatcher as a means of "exploring whether space could one day host scalable machine learning infrastructure."

The physics argument is straightforward. In low Earth orbit, satellites can "access near-constant sunlight, generating up to eight times more solar power than on Earth," Alphabet said in the post. The company eventually expects to be able to "link together multiple constellations of satellites, allowing them to manage larger AI workloads while in orbit."

That last point matters for the long-term economics. Single satellites can only carry so much computing hardware. If constellations can be networked into a distributed orbital data center, the theoretical ceiling on AI training and inference capacity in space rises substantially.

Alphabet has already tested its TPUs running AI workloads in a facility at the University of California at Davis. But ground testing can only simulate so much. The company doesn't yet know how its chips will perform in the challenging conditions of low Earth orbit, where radiation, thermal extremes, and the vacuum of space impose stresses that no terrestrial data center confronts.

Thursday's flight is designed to start answering that question with real hardware in the actual environment.

The Alphabet-SpaceX relationship

The mission also underscores how tightly the two companies are bound. Alphabet is a significant investor in SpaceX, which went public in June in a record IPO, and the businesses are close partners even as their AI divisions compete with one another. Alphabet's stake in SpaceX is currently valued at over $82 billion.

That dual dynamic — investor, partner, and competitor at once — is increasingly common across the AI infrastructure sector, where the capital requirements of compute, launch, and energy push rivals into shared ventures.

SpaceX's own orbital data center plans

SpaceX, led by Elon Musk, is not merely a launch provider in this story. The company has announced plans to build and launch its own orbital data centers. These will be comprised of swarms of satellites the company develops in Redmond, Washington, equipped with graphics processing units and solar arrays it plans to produce with Tesla, Musk's automaker.

Musk has put an aggressive timeline on the idea. Earlier this year, he said data centers in space would be the cheapest way to train AI, "and that will be true within two years, maybe three at the latest." SpaceX COO Gwynne Shotwell offered a slightly more concrete marker at an event in September, saying the company will deploy "supercompute in space" in 2027.

If those timelines hold, SpaceX would move from concept to deployed orbital computing infrastructure in roughly the same window that ground-based AI data center buildouts are projected to strain power grids in major markets.

The engineering and economic obstacles

Industry experts remain skeptical, and for concrete reasons. They say space-based data centers represent a far-out mission, if they can even become feasible at all.

Rocket launches remain capacity-constrained and expensive. Every GPU or TPU destined for orbit must first survive a ride to space, and launch cadence — even SpaceX's — is finite. Orbital data centers would also require cooling systems and chips that can withstand extreme temperatures, as well as protection from radiation, none of which are solved problems at data center scale. Clutter and orbital debris could also impede their viability, a growing concern as low Earth orbit becomes more crowded.

Cooling in particular is a fundamental constraint. Terrestrial data centers dissipate heat into air or water. In the vacuum of space, there is no medium to carry heat away, forcing reliance on radiative cooling, which is far harder to scale.

Thursday's test will not resolve any of these questions. But it will give Alphabet its first real telemetry on how TPUs behave in orbit, data that will shape whether Project Suncatcher graduates from moonshot to engineering program.

A busy day for SpaceX

The Transporter-18 flight will be the second potentially historic launch in a single day for SpaceX. The company is also slated to carry astronauts to the International Space Station from Florida's Space Coast for NASA on Thursday, marking the start of a six-month mission in orbit.

Two launches in one day — one carrying crew to the ISS, another carrying the first prototype of an orbital AI data center — capture the two trajectories SpaceX is riding simultaneously: its established government and rideshare launch business, and its push into AI infrastructure as a new frontier.

For Alphabet, the payoff is longer-term. If Project Suncatcher's satellites can demonstrate sustained AI compute in orbit, the company gains a potential path around terrestrial energy and siting constraints. If the chips fail in orbit's harsh conditions, the space data center vision loses its most credible early test case.