Forget about about wireless power transfer efficiency, the normal wireless power amplifier (PA) efficiency for cellular wireless applications (or any wireless systems) is probably just around 10 - 20%. This inefficiency is terrible not only because the use of modern linear wireless power amplifier but also due to other factors.
For starter, there is a significant distance of wired connection between PA and the antenna that resides on the tower (yes wireless PA still need wires to function). This distance create delay effect on the signal and suddenly this unaccounted impedance mismatched (because delay is phase), and changing phase results in totally different impedance on the Smith Chart even for exactly the same signal amplitude.
Secondly, wideband signal impedance matching (tens to few hundreds of MHz that are now the norm rather the exception) is impossible to do in analog domain as practiced by the industry and putting anything and everything not working as the mythical "memory effect".
Can't wait until the climate change people get into the matter and realize that about half of the wireless systems power budget is due to the inefficiency of the freaking PA and all hell breaks loose.