Nelson Pass explains that Class A amplification is very inefficient (we all know that) & that high-biased Class AB deserves more consideration, as it performs almost as accurately but consumes many times less energy.

(As I understand the gist of this article in Positive Feedback.)

" ...a highly biased AB amplifier, also referred to as A/AB. No shame in that—earlier we have seen curves showing that an amplifier that falls short of full Class A at rated power is nearly as linear. "

" Class AB amplifier with a decent bias current still offers fine performance. "


(Pass Labs X-600.5 mono amps)

DarkLantern blog - darklanternforowen.wordpress.com

Yes, for a transistor amp, as there is enough negative feedback to correct the slight difference in transconductance between class A and high-bias class AB. Denon years ago (in the POA 4400A and 6600A) had a dynamic class A bias arrangement that has low bias at small outputs and high bias at high outputs. So the amp runs much cooler but will still be in class A. Krell has a dynamic bias system as well. There also exists an IC that automatically tracks bias and current on output transistors at all times and eliminates switching distortion altogether. But the distortion in dynamic bias arrangements is still higher than with pure class A.
Class H has the lowest distortion with reasonable efficiency. The main amp is class A  at high current on very low voltage power rails to reduce heat, and its power supply is modulated by a class B amp operating with typical class B efficiency.
For a valve amp though, class A results in the lowest distortion, and is very inefficient.

AMR-iFi R&D

I was interested too, to see this general 'rule of thumb'...

" ...regular Class A amplifiers tend to weigh about a pound per watt, and they tend to cost about $100 per watt (or more). "

DarkLantern blog - darklanternforowen.wordpress.com

Agree, big heatsinks and robust output stages = weight, size and $$$! But is worth it.

AMR-iFi R&D