It gets even worse.

From this article from early 2025, we get a median life for Starlink satellites of 5.3 years, though expected to improve with newer versions of the satellites.

From other more broad analysis of the life expectancy of low orbit satellites we get that the solar cycle - which is 11 years - means that at its peak many more such satellites die due to the Earth's athmosphere expanding (so the satellites are more impacted by athmospheric drag hence are less likely to manage to remain in orbit, which they do using engines which have limited propellant). This is in addition to normal hazards of space, such as failures due to solar activity.

So expect that even for the lastest generation, half-life of these satellites is between 5 to at best 11 years.

Deutsche Telekom does not have the capital costs of replacing half its network every 5 - 11 years.

I don't disagree that even starlink is overvalued, my point was that it's the only thing in spacex bringing in profit and why.

FWIW, their v3 satellite model is apparently a 10x bandwidth increase over v2 (although weights about 5x as much). At the moment they are so oversubscribed that they can charge a $1000 surcharge for people to just sign up in some areas, so there's definitely demand. So given they're already around 10 mill subscribers there is some growth to be had (although I agree it won't be 100s of millions, maybe a few 10s - far from the 50+ multiplier on the valuation, even ignoring the other 2 money pits bundled in)

That being said, you're entirely correct that fixing the on-ground infrastructure is a much better investment if the goal is good internet service and would deliver a better quality service for cheaper, but that is happening extremely slowly in most of the developed countries (e.g. UK has terrible internet infra as well), especially in rural areas. So for the next 10 years starlink doss have a large market because the ground infrastructure sucks, and it's not "hype" to spend money on it for investors where's space stuff is.