First and foremost - In the words of Jeremy Clarkson, "Power!" Continuous power output is actually a statistical measure called the root mean square - RMS.
If you consider the graph of a sinusodial function as the output function of an audio amplifier, then the area under that function is equal to the power produced by said amplifier. As this function oscillates positive to negative, any periodic measurement would sum to zero as you would have equal positive and negative parts.
The S in RMS: The RMS measure avoids the above zero sum scenario by squaring of the output function at each point in the time domain (time domain is fancy talk for "when shit's happening") in order to guarantee that the value is positive.
The M in RMS: Since the function is a continuous wave that varies in amplitude with time, the values of all of the power measurements taken from the output function are added together and divided by the number of measurements taken over our time domain in order to achieve an average - or mean - value of the sums of the squares of the sampled values of the output function.
The R in RMS: Since all the power measurements needed to be positive in order to prevent the math telling us that the amp is putting out 0 power (again, due to sinusodial output), each of these measurements had to be squared. Since we are interested in continuous power, not continuous power squared, the square root of the mean of the sums of the squares is taken.
Root Mean Square - Literally, the square root of an average (mean) value that is calculated using data points whose values are all squared.
So you see, as an amplifier does it's thing, there is always that average output power value that is the RMS measure - A statistical measure. While RMS is all well and good, it is a very robust statistical measure and does not fare well when attempting to calculate peaks in amplifier output power.
When the load from your loudspeakers demand more current than the amplifier can supply, bad things happen.
Regarding dynamic power, it's far safer to have a solid state amplifier that is entirely too powerful for your speakers than one that has insufficient power. I said earlier that amplifier power is equal to the area under the graph of the output function of a given amplifier. When an amplifier "clips" it essentially produces an output signal that exceeds the ability of the amplifier to control, thus the wave function grows in amplitude to a point where the tops of the waves are chopped or "clipped" off flat. When the tops are clipped off of a sine wave, the shape of each wave comes to resemble a rectangle instead of series of waves, dramatically increasing the area under the output function. As a consequence, power increases to dangerous levels as does distortion. Since solid state amps distort primarily in odd harmonics, the harsh clipping sound that we've all come to know and (not) love is created. A good rule of thumb is to look for an amplifier with at least 3db of headroom; such amplifiers can double their power when the demand necessary.
Tube amps tend to clip in even harmonics. Along with that and the fact that they have relatively low NFB, tubes exhibit what some folks call "soft clipping".
Based on what I've just said, I hope it's evident that you are better off running an amp that puts out 200W RMS and 450 watts peak on a set of speakers that are rated at 60W RMS input than you are a set of speakers rated at 250W RMS input with an amplifier rated at 30W RMS output.
Of course, loudspeaker impedance and loudspeaker sensitivity play a significant role in this process, but, I feel like I've already typed enough crap that nobody wanted to read in the first place.
Guys, I'm the farthest thing on Earth from an expert on this crap, feel free to correct any errors.
