Google To Launch 4 TPUs Into Space In First Project Suncatcher Test

Google is taking its artificial intelligence (AI) ambitions beyond Earth, preparing to launch four Tensor Processing Units (TPUs) into orbit as part of the first space-based hardware test for its ambitious Project Suncatcher.

The experiment will test whether Google's AI chips and supporting infrastructure can survive the intense forces of a rocket launch and operate reliably amid radiation and extreme thermal conditions in space.

 

TL;DR

 

• Google is preparing the first orbital hardware test for Project Suncatcher, its research project exploring AI computing infrastructure in space.
• The prototype satellite will reportedly carry four TPUs and launch aboard SpaceX's Transporter-18 mission.
• Google is testing challenges surrounding radiation, cooling and launch forces.
• Elon Musk believes space could eventually account for almost all computing capacity.

Google first unveiled Project Suncatcher in November 2025 as a research initiative exploring whether solar-powered satellites carrying its AI chips could eventually form interconnected computing clusters in space.

Now, the company is ready to put its hardware to the test.

According to Google's latest Project Suncatcher update, its first prototype satellite will fly aboard SpaceX's Transporter-18 rideshare mission in collaboration with satellite company Planet.

“This first launch is about seeing what works, identifying points of failure, and applying those findings to future missions,” said Travis Beals, Senior Director, Paradigms of Intelligence at Google.

Ars Technica reports that the satellite is expected to launch on October 1 and will carry four TPUs.

However, Google isn't launching a fully operational AI data center into space yet. The experiment is designed to answer a more fundamental question: can the hardware needed to power such infrastructure reliably operate in orbit?

 

Google Will Test TPUs Against Launch Forces, Radiation And Heat

Getting AI chips into orbit presents some extreme engineering challenges.

According to Google, a rocket journey into low Earth orbit lasts around 10 minutes, during which spacecraft can experience sustained acceleration loads reaching 10 times the force of gravity.

Individual components can experience considerably higher forces.

Google has already subjected its satellite hardware to vibration testing across three axes to recreate conditions experienced during launch.

Then, there's radiation.

Solar activity and cosmic rays can interfere with electronics and potentially cause problems such as bit flips. Google tested its Trillium TPUs at the Crocker Nuclear Laboratory at UC Davis using a proton beam while the chips were running AI workloads.

The company's initial results suggest its TPUs can withstand a total ionizing radiation dose greater than what they would receive during a five-year space mission.

Cooling creates another challenge.

TPUs generate substantial amounts of heat, but the vacuum of space means engineers can't simply rely on airflow to cool them. Google is experimenting with heat pipes and radiators and has tested the technology inside a thermal vacuum chamber.

The upcoming mission will give the company real-world information about how those systems perform in orbit.

 

Google Wants To Build Connected AI Computing Clusters In Space

 

Project Suncatcher's ambitions extend far beyond a single satellite.

Google envisions future satellites carrying dozens of TPUs and operating together as clusters, allowing multiple spacecraft to process larger AI workloads.

However, that requires moving huge amounts of data between satellites.

Google is developing optical links that use lasers to transfer information between spacecraft, with the company comparing the precision required to hitting a coin-sized target from miles away while both points are moving.

Google plans another experiment involving two satellites in 2027 to test such technology.

One of the biggest attractions of taking computing infrastructure into orbit is energy.

Google says satellites in low Earth orbit could receive near-continuous sunlight, potentially generating up to eight times more solar power than equivalent systems on Earth.

This could become particularly important as growing AI workloads drive demand for more computing infrastructure and electricity.

Yet, Project Suncatcher still has substantial technical and economic hurdles to overcome, including launch costs, radiation, cooling, networking and maintaining hardware that isn't easily accessible.

 

Elon Musk Says Space Could Eventually Dominate Computing

Google isn't alone in eyeing space as a potential home for AI infrastructure.

SpaceX CEO Elon Musk, whose company is supplying the launch for Google's experiment, has been vocal about moving computing capacity beyond Earth.

Reacting to Google's latest development, Musk said, “The amount of compute in space will obviously round up to 100% of all compute,” according to India Today.

SpaceX itself has ambitions in this area, while other technology and aerospace companies are also investigating space-based computing infrastructure.

However, the economics remain far from settled.

The Register cited Gartner analyst Bill Ray as questioning the near-term practicality of orbital data centers, highlighting the enormous cost of sending computing equipment into space compared with operating infrastructure on Earth.

For now, Google's Project Suncatcher remains a research project rather than a commercial data center rollout.

As Reuters reported, the upcoming mission represents Google's first in-orbit test under the project.

Its results could help determine whether Google's much bigger vision, interconnected satellites packed with AI processors and powered by solar energy, has a realistic path from research concept to computing infrastructure.