@KenKirtland17 The only niche that would realistically benefit is compute larger than what a device can process locally, but smaller than would require a data center, and needed to be even lower latency than what Starlink already provides.
@KenKirtland17 On-orbit AI is, at its core, just compute. Compute is possible on-ground, and data centers receive a substantial economy of scale. The argument that latency is reduced is minimised by the fact that Starlink already reduces latency to fixed locations...
@MultiModalGeoff This exact spot is what always comes to my mind whenever I hear people moaning about floating bus stops. People who obviously haven't had to deal with the bus stop itself actually being the cycle lane.