[PunchyHamster]

technically we could use 40kHz and just upsample, the extra frequency over 40kHz is basically leeway to make analog part possible/cheap, but it is not technically needed in the signal

the first CD player didn't had compute power to upsample perfectly but modern devices certainly do.

[mort96]

AFAIU, 40kHz exactly wouldn't really work, if your goal is to represent 0Hz-20kHz: in order to avoid aliasing, you need a low pass filter to remove all frequency content above half your sample rate, and no filter is infinitely hard (and you generally want to give the filter a decent range of frequencies to work with). If you want to start your low pass filter at 20kHz, you want it to end (i.e reach practically -∞dB) at a few kHz above 20kHz. If you used a sample rate of exactly 40kHz, you would need your low pass filter to reach -∞dB at 20kHz, meaning it'd have to start somewhere in the audible region.

Though this is just my understanding. Maybe I'm wrong.