Well, so is satellite launch right? Cost, efficiency, and scaling are hard to do. That's SpaceX's entire raison d'etre. Doing a general public usable all weather maintenance free well designed phased array terminal they can sell for $250 and pump out by the millions is as worthy an achievement as near anything else in the Starlink project. And I'd love if it was more available too even terrestrially, for PtP/PtMP links alignment even motionless is a certain amount of work at long distances. And long range high bandwidth stuff isn't cheap. It'd be pretty cool if you could have units for $250 that you just needed to aim vaguely in the right direction and then it all just worked.
The Starlink terminal is a very cool piece of kit, but it's not nearly as interesting as what they're hucking into LEO, and how they're doing it.
The famous phrase 'Quantity has a quality of its own' comes to mind.
SpaceX was the only conceivable launch provider for this, and if it had been an external customer that cares too much about the risk of these launches the incremental improvements that made this cost-effective wouldn't have been possible. Realistically this was only viable for SpaceX doing it as part of R&D for their own rockets. And even then this puts severe financial strain on them because their original business plan was built around having Starship available years ago for even cheaper deployment of bigger satellites
Of course now that it has been done and technology has advanced by ~7 years it is much easier for new mega constellations. But at the time SpaceX started doing it the idea was rightfully called insane