Gwynne Shotwell (in case you don't trust Elon) has stated that Starlink is profitable, and Elon has said it as well

https://www.youtube.com/watch?v=bUAunM6abeY (Shotwell video)

From a profitability standpoint, it doesn't matter if they have to keep re-launching satellites every 5 years, as long as the business with launching satellites is profitable. If they spend 5 billion doing launches every year and then after all their costs have any profit they can just keep spending 5 billion every year to replenish the network and keep making whatever that profit is.

Their profit gets capped by the amount of bandwidth and customers they can find, but it is predictable as they know their bandwidth and how many customers they can support per launch. By the time they are replenishing the network, if it was simply the same satellites and launch count their growth would begin to plateau, however every 5 years they can upgrade the satellites to the latest tech, so each 5 year cycle they're actually able to increase their service capacity which means more customers and more profit. There's already a big difference in capacity between their v1 and current satellites in orbit in terms of overall bandwidth per launch, and SpaceX keeps increasing their launch cadence.

Further, if Starship is successful, they'll start offering gigabit services to consumers, and their costs would be dramatically reduced as now they have a reusable 1st AND 2nd stage. Starlink is already profitable with a disposable 2nd stage.

Just trying to compare it to fiber+5g completely misses the point that these LEO networks can provide internet to other things as well. Planes, boats, military, government services, remote areas fiber and 5g will never go, like a remote base in Antarctica. In critical setups people even use it as backup internet in fiber connected areas. All of that extra stuff is what helps make it profitable. It probably wouldn't be profitable if it was simply servicing rural communities.

You are right that the upkeep of a fiber+5g system is less, but it costs millions of dollars to support a very small amount of people, so it takes a very long time to recoup that cost, but once the cost is covered, then the profits start rolling in for a very very long time. The problem is the big telco's would rather spend less money to connect more people in urban areas and recoup that cost faster. (edit: On the other hand Starlink has to be able to recoup that cost (and more) in 5 years or the business simply doesn't work)

Starlink isn't meant to compete with this kind of service if it's installed, but the Teclos don't want to install it due to the high upfront costs and very long pay back time.

Edit: Just to be clear, they still have to make back the cost of the original network when they weren't profitable, but as long as they are profitable in its current state, they'll get there. While we don't know the exact cost for the existing network, it's probably around $8 billion dollars as we have rough ideas of launch costs + dish costs. In 2025 they're going to make over 10 billion in revenue. They haven't publicly disclosed how much of that will be profit.