Copenhagen gives up determinism, definiteness and locality. It's by far the worst of the interpretations, but chugs along due to inertia
True that.
Copenhagen can't actually explain where the quantum speedup actually comes from.
Other interpretations try to explain it, but are similarly problematic. For instance, in many worlds, Everett tried to argue that pieces of the computation are shuffled off to other worlds and then brought back, but this is problematic because information isn't supposed to be shared with other worlds, and since other worlds also shuffle off an equal amount of computation to those worlds, how is overall speedup achieved exactly?
In pilot wave theories, quantum computing has a simple story: because the world is deterministic, the entire history of time leading up to your computation was predetermined and part of the computation. "Programming" a quantum computer simply creates the conditions where you can read out the answer.
The apparent quantum speedup is actually part of an illusion that we and our equipment are separate from the rest of the universe, but quantum computation is really a classical computation that had lots of time and plenty of resources (all particles in the universe) to run.
Who cares? Experiments are a tie: those different interpretations all make the same observable claims.
From there, which is best depends on what you care about most: is the first interpretation we came up with best, because "science"? Is the simplest interpretation best, because "Bayes"? Do stuff like locality, determinism etc. matter?
Me I side with Bayes, and that means Many Worlds/decoherence, which is simplest. Others would make another choice.
You're probably thinking of "instrumentalism" aka "shut up and calculate". Copenhagen is not instrumentalism.