I have doubts it will ever be "Supreme" but I can genuinely believe it will help locate solutions, subsequently confirmed by traditional methods, which are then optimised to be as quick or quicker.

Run enough times, find the statistically satisfying minima or maxima or point of interest. Mainly it's about coherence and noise. Stable qbits are like Noah's cubits: hard to find.

ubiquitous RSA factoring in nanoseconds? Don't personally think so. Specific key finding for post-hoc analysis? Dunno, but not in 2022 or by 2030 (to agree with you)