On October 1, 2026 a SpaceX rocket carried a Google satellite with four AI chips into orbit to test whether a data center can run in space. The finale lays out what orbital-computing companies have filed and claimed, gives skeptics their full case on heat, radiation, launch cost and debris, and recalls telecom forecasts that failed. Its verdict is that the forecast's direction holds up better than its timing.
Who's who
- Travis BealsGoogle senior director who leads Project Suncatcher
- Brandon LuciaCarnegie Mellon professor quoted on the cost of computing in orbit
- Elon Musk+2 more episodesChief executive of SpaceX and Tesla
- Sundar PichaiGoogle chief executive who projected large 2026 capital spending
- Donald KesslerNASA scientist who described cascading debris collisions in orbit
- Burton Cour-PalaisNASA scientist and co-author of the 1978 debris-belt paper
Where it happens
- Bastrop, TexasSite of SpaceX's planned Gigasat satellite factory
- Low Earth orbitHome of Starlink and proposed orbital data centers
Timeline
- June 1978Kessler and Cour-Palais publish their paper on creating a debris belt
- August 13, 1999Iridium files for bankruptcy nine months after opening
- November 4, 2025Google announces Project Suncatcher
- January 2026SpaceX applies to the FCC for up to one million orbital data-center satellites
- June 12, 2026SpaceX shares begin trading on the Nasdaq
- October 1, 2026Google's first Suncatcher prototype launches on a SpaceX rideshare mission
Still debated
When, if ever, orbital computing becomes cheaper than ground computing is open: Musk says within two to three years, a Google lead says not within five, and Google's analysis points to the mid-2030s at best. Whether SpaceX's million-satellite filing is a plan or a placeholder is also unresolved.