Haven't the economies of scale already kicked in a little? Starlink is profitable. I'm not sure if spacex minus all starship costs is profitable. I'm pretty sure they're quite good at building them faster, better, and for less now. Please show me otherwise. I guess you could call that more learning curve, but that's a fuzzy line.

I guess overall I don't understand your point. What does it matter that the majority of their payload mass was their own? And nobody asking for something is certainly not the same thing as nobody wanting/needing it. See: cars, computers... Starlink now exists and I'm quite certain people love it and rely on it. Wouldn't exist without falcon 9. Starlink basically solved the last mile problem, surely you'd agree rural folks having access to Internet is a good thing?

They made their own demand for falcon 9, with that scale bringing down costs enough to raise demand for basically every research org needing satellites to contract with SpaceX.

I don't think we can definitively predict the demand either way for payload at $300/kg or whatever without first getting there, but Jevons might have some ideas.