Google’s October 1, 2026 launch of its Project Suncatcher prototype—four Trillium TPUs on a SpaceX Falcon 9 rideshare satellite—marks the first major tech company test of orbital AI hardware, yet falls well short of the 100-accelerator threshold required for market resolution. Traders price near-term outcomes low because current efforts remain small-scale experiments focused on radiation tolerance, thermal management in vacuum, and basic Gemini inference rather than production clusters. Starcloud’s prior low-Earth orbit H100 training run and upcoming lunar-orbit payload, alongside SpaceX and Blue Origin constellation filings, signal broader industry momentum powered by abundant solar energy and power-grid relief, but economic viability hinges on drastic reductions in launch costs and laser inter-satellite links. Key 2027 catalysts include Google’s planned pair of follow-on satellites and potential larger AI1 deployments, though full-scale orbital data centers face significant engineering and regulatory timelines.