Orbital AI Data Centers: Son Cites Latency and Launch Cost as SpaceX Race Heats Up

Orbital AI Data Centers: Son Cites Latency and Launch Cost as SpaceX Race Heats Up follows Masayoshi Son’s critique of putting AI compute in orbit. Speaking at SoftBank’s mobile unit annual meeting on June 23, Son—who chairs the $500 billion Stargate AI infrastructure initiative with OpenAI—told shareholders that Elon Musk’s orbital data-center plan fails basic cost tests. Son argued chips, not electricity, drive data center expenses, and that orbital systems would quickly lose on launch costs, on-orbit maintenance, and inter-satellite latency that would hinder tightly synchronized GPU training. He added that the AI race would be settled on Earth well before orbital compute matures. A TechCrunch analysis published June 27 notes that opponents including Son, OpenAI’s Sam Altman and AWS CEO Matt Garman have financial stakes in terrestrial infrastructure. The article also recounts SpaceX’s FCC filing for up to one million satellites as orbital data centers between 500 and 2,000 kilometers, and Musk’s June 8 technical briefing for the AI1 spacecraft with a 70-meter wingspan and up to 150 kilowatts peak.