To me, Starlink looks like a very typical Musk venture - vastly, vastly overhyped. As someone who would like to move to a more rural area, it appeals to me, but the numbers don't work.

The _total planned switching capacity_ of the starlink constellation, and I emphasize planned because the current capacity is kind of a joke, is approximately that of a single current generation multi-linecard network switch. And that's assuming they do in-space switching, which at the moment they do not, and to my knowledge none of the current satellites are even capable of doing so in a non-experimental production capacity. They also have very significant routing challenges if they ever do that (and even if they don't). Worse, the skinny end of the pipe is on the wrong side of the satellites, so even things like caches are not really going to help them.

It just isn't even remotely possible that Starlink is going to provide the current level of service once they get any kind of significant user uptake. The bandwidth per spot is too low and the spots are too large.

I personally think that half the Starlink story (inter-satellite switching) is going to be SpaceX's "full self driving." Unlike FSD, SpaceX has an easy out - they will build out enough ground station coverage that they don't need to do inter-satellite switching except for the air/sea cases and those markets they may simply abandon.

The true innovation of Starlink is that their ground stations are dirt cheap, but that's not a sexy story.