For obvious reasons. Almost everything can tolerate 5V. Very few things can tolerate 9V and above.
Consequently, you need a complicated dance in order to ensure that everybody agrees before jamming 9V, 15V or 20V down the pipe. That dance has enough complexity that effectively everybody uses a full-blown Cortex M0 with a couple of purpose built analog blocks to carry it out.
See: https://www.infineon.com/cms/en/product/universal-serial-bus...