SpaceX has two big things going for it: Starship/Raptor factories churning them out cheap, and themselves as a paying customer in Starlink who can dogfood a risky new launch platform.

Now that they're demoed pez-dispensing v3 dummy Starlinks, I'd assume they'll start launching real ones within ~1-3 months. At that point as long as they can deliver payload to orbit and catch the booster the program is operational and they'll start switching their own Falcon 9 launches over.

The HLS timeline is definitely dicey, but whether Starship winds up being the blocker remains to be seen. Otherwise they've now succeeded enough to "lean launch" Starship with equal/better capabilities to any other existing orbital rocket, and Starlink can fund indefinite further tests/iterations on the rest of their roadmap features (which no one else has).

> I'd assume they'll start launching real ones within ~1-3 months

I think it's still a bit early for that given that they've only had one successful flight and are still testing lots of new design changes, but I think you're right that the capabilities they've already demonstrated are probably enough to make Starship commercially viable even if literally none of the other revolutionary improvements they're working on pan out.

Falcon 9 doesn't recover the 2nd stage at all, and it's already by far the least expensive rocket out there in terms of cost/kg.