there are two notes --
DC is the average value. normally over the entire signal, but for practical use it is just the average over some "long enough" time.
the real part of the "clipping is DC" vs "clipping is AC" is based on what "long enough" means. if you look only at select parts of the wave, yes, it would be DC.
it seems unreasonable to look at less than one cycle to determine this -- after all, the speaker can play 20hz, so 1/20th of a second isn't called "DC". If one full cycle is considered, then clipping will have little to no "DC" component. This makes sense because the wave spends as much time above zero as below zero. (and equal magnitude). the DC component will be based on some difference in high vs low time or high vs low magnitude.
amplifiers can output DC, and some designs can do so more than others. but commercial grade amplifiers will typically be designed for a low DC offset. either from design, or from production tuning. amplifiers without feedback -- that don't compare the actual output to the desired output and attempt to correct -- will generally have higher DC offsets.
The real DC output case is when an amplifier blows. in that case, a LARGE DC offset is applied. (if not otherwise prevented).
DC isn't good for speakers, though at low power its ok. the speaker has the minimum amount of impedance at DC, and all power is real power. further, a DC-only signal means the speaker is held out or in, and is not moving. in that case, any forced-air cooling will not function. even in cases of some AC signal moving the cone, the speaker is still needing to dissipate more power for a given amount of movement.
