Indeed, BUT guess what happens when we take a waform of only limited time … we’ll get DISTORTION. This distortion is greatest at the very beginning and very end of the waveform.
Now let me give you a homework: calculate how many samples it takes in worst case (from beginning or end of the file) to have this distortion created by time-chopping to be lower than quatisation distortion (you can choose the sampling parameters yourself). Or you may also choose to calculate how many samples it takes to make this distortion less than … let’s say … 1%. That would be more like a real-world problem.
BTW, we DAW users all should be very familiar on problems using Shannon-Nyqvist where it’s not applicable: those clicks what we get when slicing and combining waveforms without using appropriate methods.