This is not how this works. One day in the future, when quantum computers have matured enough to do something actually useful instead of just quantum benchmarks, they still will not be general purpose systems.
The situation will be more like video cards at the moment: it would be a subsystem doing something very specialized and limited, being controlled by a driver handing over certain jobs from the OS of the real processor.
To add to what most people are saying here, i also believe that quantum computing will evolve rapidly one day, with new algorithms being developed there is no telling if quantum computing will truly stay niche or specialized and can't be expanded into general purpose computing. as tech is always evolving, i would argue that claiming that quantum processors stay akin to gpu's. gpu's are pretty much a sprecialization of the same stuff that builds a cpu, to my understanding. therefore, there is nothing truly proving that qpu's can't just evolve backwards into a parallel to cpu's. besides, i can see it being favorable to keep only a qpu for most desktop platforms and only optionally a cpu that would be plugged into the equivalent of a pcie slot. that's far fetched, but i also don't think it's too unreasonable.
Quantum circuits aren’t general-purpose computers—they’re added to conventional computers to allow them to perform a small handful of algorithms more efficiently. I don’t believe any of those algorithms would benefit the basic features of an operating system enough that it would make sense to modify an OS to require the use of one.
(Although I could totally see Microsoft doing something like only licensing their circuit’s drivers to run on Windows.)
I think quantum computers may be impossible. But if they are possible, they will be a USB/PCIe accessory that works alongside an ordinary processor running an ordinary operating system.
I expect Linux will have a driver for quantum computers before Windows.
What exists is a weird engineering experiment that runs some synthetic tests that are designed to return a number that you plot on a graph which you show your investors. And it costs all the money in the world and then some.
Not only there is no practical use for all that, there are debates about what the practical uses might even be in theory for something that nobody really sure is happening.
I suspect the Linux kernel would support quantum first. Somehow I don’t see a multi billion dollar multinational moving fast enough to beat some caffeine addicted teen looking for street cred.