Google has announced the first practical test of Project Suncatcher, its orbital data centre design intended to prove the feasibility of running large-scale AI workloads in space.
Project Suncatcher was first announced in November 2025as a potential route to running a solar-powered data centre in space running AI workloads using Google’s proprietary tensor processing units (TPUs).
The first test in a years-long journey to this goal will happen on 1 October, when Google is set to launch a prototype satellite equipped with TPUs aboard SpaceX’s Transporter-18 ‘rideshare’ mission in partnership with the satellite manufacturer Planet.
Project Suncatcher is intended to overcome land, power and grid constraints common to data centre construction. When active, each satellite will follow a sun-synchronous low Earth orbit, passing pole-to-pole at an altitude of between 600 and 800 kilometres to receive almost 24/7 sun.
Travis Beals, senior director and Project Suncatcher lead at Google, said: “One question people ask is, ‘Why not just do solar on Earth?’ But if you put a solar panel in the right orbit, it generates five to eight times more power than that same panel down on Earth.”
Through the launch, the tech giant and AI hyperscaler aims to gather valuable orbital data and discover points of weakness. In 2027 it will launch two further satellites, to run an operational test of the lasers that will link clusters together.
When fully operational, dozens of Google’s satellites will orbit the Earth together in tight clusters up to two kilometres in diameter. In an 81-satellite configuration proposed by Google, this would see each satellite manoeuvre to always keep itself 100-200 metres away from its each of its neighbours.
Individual satellites will communicate with the others in the cluster via high-bandwidth lasers, both to share computational data and to maintain relative distance.
“The technology in space already exists, but most state-of-the-art systems are optimized for low bandwidth across large distances, whereas our lasers need to operate at very high bandwidth over extremely short distances,” Beals explained.
“Maintaining the necessary connection requires extraordinary precision, similar to hitting a coin-size target from miles away while both points are in motion. We’ll test our work on this in 2027 when we put two satellites in orbit.”
Critics have argued that chips would struggle to operate reliably in space, due to the potentially catastrophic damage space radiation can have on electronics.
Google said it ran tests on its sixth-generation Trillium TPUs within a proton beam facility at UC Davis’s Crocker Nuclear Laboratory, firing protons at the chip while it ran AI workloads. The results showed the TPUs were resilient enough to operate effectively across a five-year orbit.
Another commonly cited bottleneck for space-based data centres is heat dissipation. AI chips such as Google’s TPUs generate a large amount of heat when in use, which is typically addressed through conduction or convection methods such as water and air cooling.
Because the vacuum of space does not allow for these conventional cooling methods, satellites have to vent heat using radiators, which emit the heat from pumped coolant as thermal radiation. Google said it has run tests in a vacuum chamber but will need to gather more data on cooling through launch tests.
The first Suncatcher mission is designed to gather in-orbit data and identify potential failure points, rather than demonstrate an operational orbital data center, Google said.
It aims to launch two satellites in 2027 to test the high-bandwidth laser links needed to connect future computing clusters.
