SpaceX is set to loft Google tensor processing units into low Earth orbit on Thursday as the first flight test for Alphabet’s Project Suncatcher, a moonshot effort to study whether solar-powered AI computing can work in space. CNBC reported Oct. 1, 2026 that a Falcon 9 Transporter-18 rideshare from Vandenberg, timed around 11:15 a.m. PT, will carry Planet Labs satellites including a solar-powered prototype hosting Google TPUs.
What Project Suncatcher aims to prove
Alphabet first described Project Suncatcher in November 2025 as a research program exploring scalable machine-learning infrastructure in orbit. Google has already run TPU AI workloads in a University of California, Davis facility, but has not yet measured how the chips handle launch vibration, radiation, thermal extremes, and vacuum cooling in LEO. The company has said satellites can access near-constant sunlight and generate up to eight times more solar power than on Earth, with a longer-term goal of linking constellations for larger orbital workloads.
SpaceX partnership and competing orbital compute plans
CNBC noted Alphabet’s stake in SpaceX is valued at more than $82 billion after SpaceX’s June IPO, even as the companies’ AI efforts compete. SpaceX and Elon Musk have separately promoted orbital data centers; SpaceX COO Gwynne Shotwell said in September the company expects to deploy “supercompute in space” in 2027. The Transporter-18 flight is also part of a busy Thursday for SpaceX, which was carrying astronauts to the International Space Station from Florida the same day.
For more on SpaceX’s recent orbital milestones, see our explainers on Starship Flight 14’s first orbit and the earlier preview of Flight 14’s launch plans.
Why experts still call space data centers a long shot
Industry analysts told CNBC that capacity-constrained, expensive launches, radiation hardening, cooling, and orbital debris all remain major hurdles. Thursday’s mission is framed as an in-orbit hardware test, not a full orbital data center. Even so, the flight marks a concrete step in the race to move AI infrastructure off the ground as terrestrial data-center buildouts face power and community pushback—themes that also surface in coverage of Microsoft’s Quine biology AI push.