Alphabet’s Google said Thursday it will launch a prototype satellite next week in the first in-orbit test of Project Suncatcher, a research effort exploring whether space could eventually host large-scale AI computing infrastructure.
The mission, scheduled to fly on SpaceX’s upcoming Transporter-18 rideshare launch, is being carried out in partnership with satellite imaging company Planet Labs.
The prototype will carry Google’s Tensor Processing Units into low Earth orbit to test whether the chips can operate reliably outside Earth’s atmosphere.
What the mission will test
The satellite will measure how Google’s AI hardware withstands the forces of launch, radiation exposure, and extreme temperature swings in orbit, according to Google.
A trip to orbit takes about ten minutes and can subject onboard equipment to forces of 50 to 100 times Earth’s gravity, according to a separate report on the mission.
Google previously tested its Trillium-generation TPUs by running AI workloads inside a 67 MeV proton beam at UC Davis’s Crocker Nuclear Laboratory, simulating the radiation exposure a satellite would face over roughly five years in orbit.
The chips showed no hard failures up to the maximum tested dose of 15 krad(Si), said Google.
The company said the most radiation-sensitive component was High Bandwidth Memory, the stacked memory modules that feed data to the processing cores.
Its testing suggested the issue remains manageable within the mission’s expected exposure.
Airflow cooling is impossible in the vacuum of space, so Google is testing a different design.
It combines heat pipes with radiators, an approach it says has worked in ground-based vacuum chamber testing so far.
Why Google is doing this
Google first announced Project Suncatcher in November 2025.
The company framed it as a long-term research moonshot addressing one of AI’s most persistent bottlenecks: the enormous power demands of training and running large models.
Those demands are increasingly running up against terrestrial power grid limits.
Google says solar panels in low Earth orbit can generate up to eight times more power than equivalent panels on the ground, since satellites in the right orbit receive near-constant sunlight.
“Like any moonshot, it’s going to require us to solve a lot of complex engineering challenges,” Google CEO Sundar Pichai has said of the project.
Google’s research paper outlines a longer-term vision of compact satellite constellations.
Clusters of up to 81 satellites, spaced within a 1-kilometer radius and connected by free-space optical links rather than radio, could help overcome the signal losses that typically limit long-range space communication.
Early tests of that optical link technology have already achieved 800 Gbps transmission speeds each way, according to Google’s paper.
Not the only player racing to space
Google is not alone in pursuing orbital AI computing.
Nvidia-backed startup Starcloud launched an H100 chip into orbit in November and has already used it to run Google’s own Gemma AI model.
SpaceX is also pursuing its own data center ambitions in low Earth orbit.
Experts have cautioned that space-based AI computing remains years from commercial viability, given high launch costs, engineering constraints, and satellite production bottlenecks.
Google itself has said this first mission is designed to gather in-orbit data and identify potential failure points, not to demonstrate an operational data center.
The company plans to launch two satellites in 2027 to test the high-bandwidth laser links that future computing clusters would need.
Alphabet’s Class A shares (GOOGL) were trading at $339.28 on Thursday, up 0.43%.Thursday’s move follows Wednesday’s 3.8% decline.
The post Google to launch its first AI chip into space next week under Project Suncatcher appeared first on Invezz