As far as I understood, I always thought that they just used other people’s rockets to piggyback their own satellites to space basically for free

An analysis from early 2025 showed that the median life expectancy of a 1st generation Starlink satellite was 5.3 years, though expected to improve in v2 and v3.

Beyond that, the solar lifecycle means that every 11 years the earth's orbit expands increasing drag on low-orbit satellites (which brings them down) as well as bombarding them with more radiation (increasing the likelihood of failures).

Other sources I've found in searching for it give their median life-expectancy as being 5-7 years.

Land-based telecom operators don't have to replace pretty much all of their transmission infrastructure - worse, in a costly to access location - more than once a decade.

Space-based tele-communications is more than proven as a viable business model, what's nowhere near proven (or anywhere close to being mathematically demonstrable) is one single operator of those being worth more than many major land-based telecom operators each with many times the number of customers of that space-based telecoms business and providing much faster network access (somebody else gave Deutsche Telecomm as an example) put together.

(How exactly in terms of actual PHYSICS will Starlink deliver via radio waves to for example all 245 million European households - necessary to justify such valuations - a reliable 1Gbps connection for €10/month needed to to beat the land operators?)

SpaceX's IPO valuation which itself is mostly anchored on Starlink, is totally unjustified and unjustifiable by fundamentals.

Its actually a great business model as they need to do actual test flights to build better rockets and instead of that being pure R&D expenses, its supplemented by launching Starlink satellites which generate their own revenue for the company.