Unless we get lucky, quantum computers aren't going to significantly increase our overall computing power. They won't "break all encryption" either; the development and standardization of quantum-resistant cryptography has moved quickly and surely, and my guess is that implementation bugs, paper-thin software infrastructures, and the unquenchable human drive to click on random links will continue to be much more dangerous than any technothriller code-breaking machine.
Where quantum computers are working very well is on the simulation of quantum systems, which is iffy as a benchmark but extremely helpful in practical terms. I expect we're underestimating the speed of future advances in sensors and material technology due to advances in analysis and engineering at the atomic scale - the sort of Industrial Sci-Fi that isn't packaged in specific devices but rather improves manufacturing, medical devices, infrastructure, etc, across the board. I would be very surprised if in twenty years there are desktop quantum computers in the most literal sense of the term, but I'd also be very surprised if in twenty years there isn't widespread use of materials that don't exist now, and if the resolution, cost effectiveness, and energy efficiency of sensors hasn't improved drastically.
Think Bayer or 3M, not Apple, and remember that the Haber-Bosch process changed the world as deeply as any other technology, ever.
Of those things I'm fairly certain. But I have also a vague feeling — hence this throwaway note — that advances in practical quantum computing will have an outsized effect on what people believe computers do and can do, and therefore on how they are used. Everybody talks now about "algorithms" as if no such thing had existed before Facebook, and the rise, rise, and fall of cryptocurrencies imbued cryptographic protocols with an aura of positively mystical financial and social power. Both things absurd, and most of the social currency of those technologies came after the success of their notorious applications, but there's inertia here — Facebook is Meta and Meta is sliding into irrelevance hounded by regulators and flighty advertisers smelling its demographic and cultural stalling — but "the algorithm" remains.
Right now quantum computers don't have that sort of impact, but I suspect that their first success with widespread implications will change that. Not an undeniable but esoteric proof of quantum supremacy in some abstract setup of deep mathematical importance, but the first revolutionary medical sensor or impossible material in a mass product, vehicle or building. Once that happens, "quantum computing" will get a renewed sociopolitical charge, which will have the interesting side effect of things that are possible now will begin to be considered possible (or rather impossible in a now possible way) - we never needed a blockchain to have distributed consensus databases, but after bitcoin prices rose enough to become a mainstream topic, everybody started looking for ways to use them. We have, today, the mathematical tool to do all sort of "quantum things" for, e.g., zero-knowledge proofs, computing across infinite simultaneous hypothesis (which is standard Bayesian statistics, but people get bored with that and yet feel they instantly understand the image of a quantum computer "computing in parallel in multiple universes" whatever that means), etc. Even sufficiently precise chemistry can be labeled "quantum" if you want, and it will. As soon as we have the first great quantum-anything success, "quantum-" will be the new "crypto-" or the new "cyber-" (remember "cyber-" ?).
There'll be a lot of bull, smoke, and mirrors, but it'll also be a window of opportunity to get some perfectly feasible but weird-looking things done. The first impact is by far the most important one (cryptography, solid quantum engineering), the second wave is questionable (bitcoin), and the third one tends to be nonsensical, but the side cultural side effects after the bubbles finish popping is often where some of the deeper influences remain, as a culture integrates to its lexicon of commonsense plausibilities the misunderstood, mixed-up remains of Ph.D. dissertations and vacuous white papers.