If you were simulating the universe it certainly doesn't help performance having to simulate an exponential quantity of interfering states rather than a single classical one. It's anything but lazy evaluation.
I remember when the "proceduraly generated" games first showed up - 30 years ago?
You could get a billion different scenarios, each one involving a million locations.
None of them existed before you started playing the game. And generating all of them in advance would be a) impossible, given the memory requirements and b) a huge waste, considering that 99.99999% of the players never needed to experience 99.99999% of the scenarios.
So what they did, they only presented up what is "observable" to you, the player.
That's not an answer to the concern. Classical mechanics has no foundation -- it's a large castle floating in the air. Quantum mechanics tries to start from much less less assumed "something from nothing structure", which is puts much more complexity in sight (instead of just hiding it in axioms) that needs to be managed somehow.
If you want purely high-performance model of the Universe, nothing performs better than "everything you observe is simply invented the moment you invent it", be it classical or quantum.
You could get a billion different scenarios, each one involving a million locations.
None of them existed before you started playing the game. And generating all of them in advance would be a) impossible, given the memory requirements and b) a huge waste, considering that 99.99999% of the players never needed to experience 99.99999% of the scenarios.
So what they did, they only presented up what is "observable" to you, the player.
So there is some saving there.