Because SpceX isn't making money off selling Internet, it's posturing (as is Amazon) for the multi-trillion dollar Golden Dome program.

In some sense the geographical concentration problems are worse with LEO satellite because LEO satellite has to cover most of the world which is mostly ocean, mostly uninhabitable, mostly uninhabited. Most of the market though is people who live in suburban and urban areas that are difficult to serve with that technology and where you are going to face pricing pressures because you have cable, fiber, DSL and wireless alternatives.

Even in the long term relatively difficult rural customers are going to get fiber and erode the market. My situation is a bit difficult because my power lines go through a wet meadow that was beaver habitat [1] but I am told they will be getting me connected in the next few months and I can ditch my fiber-to-the-node.

I think the bull case for Starlink is they get Starship working and that improves their cost structure. On the other hand, the rest of the world doesn't trust Musk and there is going to be a lot of surplus competition which is politically motivated just as there is for GPS except the economics will brutal.

I've read that China badly wants its own Starlink but it's having a hard time penciling it out because: (i) the world doesn't trust China any more than it trusts the US, (ii) China doesn't have the "Atlas Shrugged" underinvestment we have so there is a lot more fiber in places that are moderately rural [2] and (iii) China has vast amounts of ultra-rural territory but there are only so many people that live there.

[1] ... before they cut down all the trees and presumably, like any other loggers, blamed the environmentalists for shutting them down

[2] ... not to mention China developed many compact villages with moderate-large apartment buildings in many rural areas that are easy to serve