Alphabet to Test AI Chips in Orbit Via SpaceX as Google Pushes Toward Space-Based...

Alphabet is preparing to put its artificial intelligence chips into orbit on Thursday, using a SpaceX Falcon 9 rocket to conduct the first in-orbit test of a project aimed at turning space into a potential new location for large-scale AI computing.

The launch, scheduled for 11:15 a.m. PT from Vandenberg Space Force Base in California, will carry Planet Labs satellites aboard SpaceX's uncrewed Transporter-18 mission. One of the satellites will be a solar-powered prototype equipped with Google's tensor processing units, or TPUs.

The mission marks the first orbital test of Project Suncatcher, an initiative Alphabet unveiled in November 2025 to investigate whether solar-powered computing infrastructure could eventually support AI workloads in space.

Alphabet described the project as an effort to determine "whether space could one day host scalable machine learning infrastructure."

The idea marks a significant departure from the conventional data-center model, in which AI companies build large facilities on the ground and compete for electricity, land, cooling capacity and connections to already-strained power grids.

Google is instead exploring whether some of those constraints could be addressed by moving the computing infrastructure above the atmosphere.

Low Earth orbit offers one potential advantage that has attracted growing interest from the technology and space industries: abundant sunlight.

Alphabet said satellites in low Earth orbit can access near-constant sunlight and generate up to eight times more solar power than equivalent systems on Earth. The company eventually envisions linking multiple satellite constellations together so they can collectively handle larger AI workloads in orbit.

That would turn individual satellites into components of a distributed computing network rather than simply communications or observation platforms.

The concept remains highly experimental. Alphabet has already tested its TPUs running AI workloads at a facility at the University of California, Davis, but Thursday's mission will provide the first opportunity to observe how the chips operate in the substantially harsher environment of low Earth orbit.

The test will matter much to the AI industry because conventional data centers rely on extensive infrastructure to maintain stable operating conditions. Space removes many of those physical advantages while introducing different problems, including radiation, extreme temperature variations, and the difficulty of dissipating heat.

AI chips generate substantial amounts of heat, meaning a space-based data center would need specialized cooling systems that can function without the conventional cooling infrastructure available on Earth.

The economics present another hurdle. Launch capacity remains limited and expensive, meaning that putting large amounts of computing hardware into orbit currently carries a significant transportation cost. Orbital debris and congestion could add another operational risk as more satellites are deployed.

Those constraints make space-based AI computing a long-term proposition rather than an immediate alternative to terrestrial data centers. But the potential payoff is large enough for major technology and space companies to begin testing the concept.

Alphabet and SpaceX Converge on Orbital Computing

SpaceX has been one of the most vocal proponents of space-based computing, arguing that orbit offers effectively "infinite real estate" at a time when AI companies are facing growing opposition to large data-center developments on Earth.

The company is also developing its own orbital data-center concept. SpaceX plans to deploy swarms of satellites equipped with graphics processing units and solar arrays, with the hardware expected to be manufactured in part through Tesla, Elon Musk's electric-vehicle company.

Musk said earlier this year that space-based data centers would eventually become the cheapest way to train AI and argued that this could happen within two or three years. SpaceX's historic IPO earlier this year was largely based on that.

SpaceX President and Chief Operating Officer Gwynne Shotwell said at an event in September that the company expects to deploy "supercompute in space" in 2027.

Alphabet and SpaceX therefore have overlapping ambitions even as their technology businesses compete in AI. Alphabet is also a significant investor in SpaceX, with its stake currently valued at more than $82 billion. That relationship gives Google an unusually close connection to the launch infrastructure required to test a technology that could eventually compete with conventional terrestrial data centers.

The economics are nevertheless far from established. A space-based computing network would have to overcome the cost of launching hardware, replacing failed satellites, maintaining communications links, and protecting processors from radiation. It would also need to demonstrate that the advantages of solar power and orbital real estate outweigh those costs.

For AI companies, the attraction is the scale of the infrastructure challenge. The rapid expansion of generative AI has created extraordinary demand for electricity and computing capacity, forcing companies to commit billions of dollars to new data centers and power infrastructure.

Moving some workloads into orbit could theoretically provide access to continuous solar energy while avoiding some of the land-use, permitting and grid constraints increasingly affecting data-center construction on Earth.

But that does not mean space eliminates the infrastructure problem. It only changes its nature.

Instead of transmission lines and local permitting, operators would have to manage launch capacity, orbital traffic, communications, thermal management, radiation protection and satellite replacement.

Against that backdrop, Thursday's mission is seen as less a demonstration that orbital data centers are ready for commercial deployment than a test of whether one of the fundamental components, the AI processor, can operate effectively in space.

If Google's TPUs perform reliably, the company will have taken an important first step toward validating the technical foundations of Project Suncatcher. Further progress would still require proving that large numbers of computing satellites can be launched, connected, cooled, and maintained at a cost competitive with terrestrial facilities.

The timing is notable because AI infrastructure is becoming one of the industry's biggest constraints. Companies are racing to build larger clusters, while electricity shortages, permitting delays and community resistance increasingly complicate expansion.

Space offers an extreme solution to those problems, but it introduces an equally extreme set of engineering and economic challenges.

The Transporter-18 mission will also be a particularly busy day for SpaceX. The company is scheduled to conduct a separate launch from Florida's Space Coast carrying astronauts to the International Space Station for NASA, beginning a six-month mission.