All of the above answers are good. But there is one other factor that is important when the output power level of an audio amplifier is specified and that is the quality of the output signal at the specified level.
If you want just raw power and don't care about the quality of the sound, then most audio amplifiers can generate a lot more power than their specs would imply. But if you throw in a decent response curve, a low level of distortion, and a low noise level, then it becomes a lot more difficult to generate the power levels that a simple analysis of the Voltage and current levels that are possible in the output stages. Any high quality amplifier will have a very flat response curve, very low levels of distortion, and very low noise levels. And even mediocre ones will have fairly good specs in these areas. And all of this comes at the sacrifice of at least some of the available power.
In any case, a manufacturer of a high quality audio amplifier would measure the output power under circumstances where those quality factors were within certain, rather low limits. So it is not just the amount of power that the transistors can operate at nor the amount of power that is consumed in operating that final stage but rather the amount of power where the manufacturer is willing to guarantee the stated performance.
If you are comparing the power output of two or more audio amplifiers, then it is important to be sure that these other factors are taken into account. If you are willing to allow the distortion to reach high levels, then I can show you a real cheap amplifier that will deliver more power to the speakers or to a dummy load than any multi-thousand dollar, high quality one. It's like comparing horse droppings to apples.
Of course, with most listeners of popular music all of this is a totally wasted effort. I can remember my brother deliberately mis-tuning his radio to increase the distortion of the audio because he liked it that way. That was before the days when the tuner in most radios was designed to automatically lock on to the center frequency of the station. I wonder what he does now but I won't be asking him.