The amp is designed to be an ideal voltage source, and unlike Tripath, we are impervious to speaker load (explain more later). While we can argue about the doubling of power when impedance drops to 2-ohm, we are still talking about measurement using a big load resistor and compare raw voltages. Much like comparing Horsepower without the right context (a 500HP Caterpillar may not run like a 100HP Honda Civic).
For example, without a constant switching frequency and a robust closed-loop response, the Tripath technology will have variable conditions at different frequencies, plus the interaction of impedance at different frequencies, thus resulted in a bit of mess (The Q is all over the place). In NuForce design, the closed-loop response makes sure that at the end of every cycle, the net error is nulled out at a speed of 1Mhz, practically nulls the variance at the speaker terminal, so we behaves closer to ideal voltage source at much higher frequency without running into instability. In conventional Class-D design, you'll notice that many of them open-loop, and have severe phase shift so it has to be strongly filtered at 20Khz to prevent instability.
For more detail, please examine the comparison between various technologies posted in nuforce.com
In the raw power department, we simply put in an off-the-shelf SMPS that is current limited to 100W. So regardless of 8, 4 or 2 ohm, the SMPS will supply 100W of current juice before the current limit kicks in and shuts down the supply. But during the instantaneous peak, we can swing a doubling of power when the speaker impedance halves. If we have a limitless power supply obviously, we can doubling up the power all the way (up to 1152W@2 ohm peak). In fact, one of our OEM application we are driving a pure capacitive load (which is why NuForce amp works exceptionally well with ribbon, and ESL). With our reference speaker, we have been playing music without clipping, with peaks of 300W+ from the 100W amp.
