Google to Launch First AI Chips Into Space Aboard SpaceX Rocket

Google is launching its first artificial intelligence chips into low Earth orbit next week aboard a SpaceX Transporter-18 rocket to see if they can survive the journey. The prototype satellite, developed in partnership with aerospace company Planet Labs, carries four of Google's Tensor Processing Units (TPUs) to test their resilience against launch vibrations, cosmic radiation, and the thermal extremes of space. "This first launch is about seeing what works, identifying points of failure, and applying those findings to future missions," Google stated.[2][3]

The mission is the first orbital test for Project Suncatcher, Google's long-term research effort to determine if energy-hungry AI data centers could eventually be relocated off-planet. AI infrastructure is currently constrained by terrestrial electricity grids and cooling water availability. By moving compute into orbit, Google hopes to tap into near-constant sunlight, which can generate up to eight times more solar power than identical panels on Earth.[2][6]

Before a chip can process data in space, it has to survive getting there. During the roughly 10-minute rocket ride into low Earth orbit, the spacecraft will endure intense vibrations and sustained acceleration up to 10 times Earth's gravity. Individual components, including the delicate TPU chips, can briefly experience forces 50 to 100 times stronger than gravity, requiring specialized mounting and reinforcement to prevent physical shattering.[2][3][6]

Once in orbit, the hardware faces a constant barrage of cosmic rays and solar radiation that can damage electronics or cause data errors known as bit flips. To prepare for this, Google previously bombarded its Trillium TPUs with a proton beam at UC Davis's Crocker Nuclear Laboratory. The engineers monitored the hardware while it actively ran AI workloads, proving the chips could survive a cumulative radiation dose greater than what they would encounter during a full five-year orbital mission.[1][2][3]

The vacuum of space presents an unexpected thermal challenge: without air, traditional convective cooling is impossible, meaning the intense heat generated by the TPUs has nowhere to go. To prevent the processors from melting, the prototype relies entirely on radiative surfaces to expel thermal energy into the void. "We're working on a number of different approaches for this, including a combination of heat pipes and radiators to cool the chips," the company noted, adding that the system was validated in thermal vacuum chambers on Earth.[2][3][6]

If this initial hardware test succeeds, Google plans to scale the experiment significantly. The company intends to launch two satellites in 2027 to evaluate the high-bandwidth laser communications needed to link orbital hardware. Most existing space lasers are built for low bandwidth over long distances, but Google's design requires high bandwidth at short range—a precision challenge the company compares to hitting a coin-sized target from miles away while both objects are moving.[2][3]

The ultimate architecture for Project Suncatcher envisions clusters of 81 satellites flying in tight formation within a one-kilometer radius at an altitude of roughly 650 kilometers. Operating together, these clusters would act as a single orbital supercomputer. However, the company estimates that launch costs must fall below $200 per kilogram—a milestone projected for the mid-2030s—before space-based compute can compete financially with the energy costs of terrestrial data centers.[3][6]

Google is not the only company looking upward to solve the AI energy crunch. Nvidia-backed startup Starcloud launched an H100 GPU into orbit in November 2025, and SpaceX is developing its own Starmind AI satellites targeted for late 2027. As terrestrial power grids struggle to keep pace with the demands of artificial intelligence, the race to build the first extraterrestrial data center is officially underway.[2][3]