The simple answer is that some people don't agree with your hypothesis, me included. Maybe we are wrong, but that's the answer. If we all start with the assumption that class D is perfect, the discussion becomes entirely meaningless, except to a marketing manager trying to decide when old habits will die. IMNSHO we are not near that point, although Class D has certainly overcome many of its early teething issues. The NAD 3020D is pretty darn good - and that's actually a more meaningful accomplishment.
For the record, i know a fair amount about GaN, and am friends with some designer and fabrication engineers involved in it. Yes, its speed is useful in class-D, but the fundamental challenge remains the perceived sound of the reconstruction filter, which is very complex since the speaker is essentially an unknown to the designer.
I also do not feel that ANY design requires even $10k, unless one has some very special performance needs. The funny thing is that in my own designs, the device selected is one of the least contributing choices. I have used ICs, solid state and (to a lesser degree) tubes and achieved sonically very similar results - again with obvious caveats for tubes and tough loads.
G
For the record, i know a fair amount about GaN, and am friends with some designer and fabrication engineers involved in it. Yes, its speed is useful in class-D, but the fundamental challenge remains the perceived sound of the reconstruction filter, which is very complex since the speaker is essentially an unknown to the designer.
I also do not feel that ANY design requires even $10k, unless one has some very special performance needs. The funny thing is that in my own designs, the device selected is one of the least contributing choices. I have used ICs, solid state and (to a lesser degree) tubes and achieved sonically very similar results - again with obvious caveats for tubes and tough loads.
G