Going fully reusable may change that equation, but first stage reuse probably isn’t enough to make the program even close to worthwhile.
Questions like rate of successful launches, amount of payload sacrificed, etc really change the economics of the whole process. If the total payload is worth 1 billion a moderately more expense launch with higher success rate can be cheaper.
It's pretty clear there is barely enough commercial launch demand for falcon 9 (it already has ~100% of the non-foreign launch market, and there isn't a huge amount of price elasticity), so no reason at all to develop starship, apart from humans in space.
Mars is at best a long term goal for Starship, and far more likely to be just a nice story that Musk uses to motivate his engineers and investors.
It’s a terrible design for anything else, because it can barely get beyond LEO without in-orbit refuelling.
None of the competing rockets (e.g. New Glenn) resemble Starship in the slightest, because none of them are intended to fly to Mars.
Given SpaceX's business, it seems safe to assume that Starlink was a design goal with at least a similar priority to the Mars goal.
Another use case it'd work fabulous for would be a LEO space hotel business.
Finally, it's also a great rocket for any use case that involves returning large masses, even if the return is from higher than LEO. Yes, it'll be a thirsty beast requiring many refuelling trips, but the tyranny of the rocket equation makes it hard to do any better. If you want to return dozens of tons from the moon or elsewhere you'd be hard pressed to do better than Starship.
For most purposes the customer does not care one iota about the booster, they are interested in the cost per kg to get where they are going. For low orbit hydrolox imposes more handling nightmare costs than it saves in amount of rocket, it is not the fuel of choice unless you're trying to impress.
(Now, things change considerably when you looking at deep space. But methalox or even kerolox fueled in orbit still beats hydrolox fueled on the ground. And hydrolox is much less storable--your rocket costs weight, necessary to reach orbit but once you're up there a smaller engine means less wasted mass. The only advantage to a bigger engine is Oberth and that is only truly relevant if you either care about time (Apollo took an inefficient path for this reason), or because you are going to carry velocity into deep space. Look at the flight path of the Webb. The booster flew higher than the maximum efficiency path because the deep space stage was puny. It wasn't powerful enough to circularize normally, the telescope fell back quite a bit before the engine had built up enough velocity to stay up. But it was worth wasting some energy on that in order to not lift as much engine away from the Earth.)
Falcon Heavy put a car in a trans-Martian orbit, and Musk has been about Starship-like things going to Mars before SpaceX managed to launch the Falcon 1, let alone them getting a chance to bid for the return-to-Moon mission.
But the Artemis mission isn't really about doing things sensibly, it's about pork barrels. You can tell by looking at the wild disparity between the vehicles, where there's this complex process to put a handful of astronauts on a space station and transfer them to a landing vehicle… but the Lunar Gateway is smaller than Starship, and I think small enough you could fit all the parts of the Lunar Gateway inside the payload volume of one Starship.
If the USA wants to go to the moon for its own sake, they could do it cheapest by just paying SpaceX for a ride, not all the other contractors.
Settlement on Mars is out of one gravity well into another, so it's not clear if it's the best first location of a extraterrestrial human territory - Moon might be easier and more reasonable.
So the camp is split between Moon and Mars, and Musk has to be on Mars.
There is zero chance of building a self-sustaining base on either within the next fifty years, and probably within the next century.
It's not a freight problem, it's an ecology problem. Designing a life support system that is stable and self-correcting and isn't in danger of running out of some essential raw material or element isn't just an unsolved problem, it's a barely considered problem.
Ironically - or perhaps not - it would be much easier to create a self-sustaining population of machines on Mars and/or the Moon than any project that relies on incredibly complex and messy human biochemistry.
What sort of technologies become enabled when putting things in space gets cheaper?
I suspect that Musks desire to have everything reusable has severely eaten into those margins though. I personally think he'd have been better off making only the first stage ('booster') reusable for the first few years, which then lets you develop more things in parallel later (the first landers can be on mars whilst you're still figuring out second stage reusability)