So that you can flip over in an uncontrolled wheelie at an even lower fraction of the throttle? Even if there was infinite energy throughput (aka power) at zero mass, the main limiter for power per total system mass would still be the battery. In any practical setup, even in super short runtime designs, getting, say, twice the power would not all that dramatic a runtime hit if it was achieved by scaling the same motor technology and paying for the extra mass with a little battery capacity. Unless of course you want to actually use that power increase for any meaningful fraction of the runtime, then you'll obviously drain the battery fast. But a zero-mass power increase would not change that a lot either.

Increasing power density (of the motor) just isn't worth much when it does not happen to coincide with an increase in efficiency (and then the battery mass saved for achieving the same range will quite literally outweigh the mass saved by a smaller engine for achieving the same power)

The good news is that those striving for power density aren't really at liberty to completely ignore efficiency in the process because cooling is a key issue for them.