An orbit is basically real estate and probably should be treated as such.
The product they sell is a highly complex product that took half a decade of development and substantial upfront investment in technology and organization.
The difference here is that SES has to get up front investment to buy productive asset that could then make money. The never had to do a huge up front investment in technology, including a whole butt-load of very complex very unique hardware and software.
And that's not enough, because the terminals are just as big a part of having a functional LEO platform as the sats are. And that again requires a very complex development and a very large upfront investment in production.
Literally non of this stuff, neither the rocket, nor the sats, nor the operations, nor the terminals are a commodity in any sense.
You can go to a space company and ask to buy some propulsion for such a sat, no problem. But if you tell them 'we would like to buy 5000' your gone get back blank stares. Nobody just has something like that just laying around. And then you go down the line of every single part on the sat, no provider actually has the capabilities you need, and your gone have to get involved with financing all of that at least partially.
But if the board decided to stick to Geostationary orbit, they were doing the smart thing by treating it like a real estate business. Because it is literally real estate.
In the terrestrial real-estate analogy, this seems to me like missing out on potential returns because you built the wrong type of building on the land you own.
Anyway, Starlink wants to scale up to 42,000 satellites. I can't imagine that's anywhere close to the maximum number that could fit in their altitude range. After all, if arranged in a grid, that's just ~ 200 x 200 satellites. The earth's surface area is many orders of magnitude larger than that, and the satellites are occupying a larger (and much thicker) orbital shell.