Google is set to launch a prototype satellite next week as a part of its Project Suncatcher, aimed at studying the performance of its tensor processing units (TPUs) in space. This initial test will examine the potential for space to host scalable AI computing infrastructure, leveraging the constant sunlight available in low Earth orbit.
Travis Beals, Google’s senior director for Paradigms of Intelligence, highlighted that successful operation in space could enable satellites to handle larger AI workloads through interconnected constellations. The upcoming Transporter-18 mission, in collaboration with SpaceX and Planet, is designed to gather in-orbit data on TPU resilience against the harsh conditions of space, such as radiation and extreme temperatures.
Beals discussed the vigorous testing TPUs have undergone, including exposure to intense vibrations simulating a rocket launch and trials in a proton beam facility, showing promising results. The overarching goal is to refine hardware capabilities gradually through methodical validations, which could eventually position space as a viable environment for large-scale AI computation.
Additionally, Google recently announced significant investments in Finland’s digital infrastructure, further solidifying its commitment to advancing technology.
