Yes. However beware that when measuring peaks a lot of test kit either effectively or explicitly averages what it measures over some time window. If so, true peaks are going to be higher than the measurement. This is especially so for SPL meters. For example, see here for John Atkinson measuring true audio peaks in a concert hall.
It may not matter that over an hour's music a small number of instantaneous peaks get clipped. At some point clipping will become audible, but blanket statements on amplifier power are indeed not credible.
This was done using an oscilloscope, where you might expect a channel bandwidth much greater than the audio bandwidth, and (in my case at least) with the brilliance wound up on the CRT so short duration peaks can be seen.
Actually I do recall a fairly recent Harbeth video of a demo where Alan Shaw was present, of a dealer demonstrating peak levels in the several hundreds of watts. This was with purely electronic music. I guess this could potentially give a very much greater crest factor than any natural acoustical music source, e.g. delta functions are not a problem for an electronic generator, but very difficult to produce on a violin or to sing!