While people with MPE/ch-per-note controllers are rare, I would assume people that would buy an 8 or 16 voice synth might also expect it to be more than 2Xtimbral.
The SCI Six-Trak is 6Xtimbral
The DSI Prophet 08 is 2Xtimbral
The DSI Tetra is 4xtimbral.
The DSI Prophet Rev2 is, most likely, also only 2Xtimbral.
Of course, building patches and then assigning the same voice to a different MIDI channel to make a ch-per-voice multi is a PITA, but it *works*. Editing the multi after its created requires editing each individual voice identically, but once a multi is built, performance works pretty great, in my opinion.
Hypothesizing here, but if 8-16Xtimbrality was possible hardware-wise, it might just be a software improvement to make easier MPE patch building/editing. Specifically, while in multi/combo/ch-per-voice/MPE mode making an edit to the voice would make the edits to ALL voices in the multi/combo. Currently with a multi, if you edit a patch at all, it only changes the copy of the patch that was put into one of the voice slots. This exposes the need for two different types of combo/multi modes. I'm sure when DSI multi-mode/combos were developed, the idea was that each voice in a combo could be intentionally different so each voice could be used for different sounds.
If there was a second combo/multi mode specifically for MPE/ch-per-note, a user could assign a patch to ALL of a combo, rather than cloning patches to each slot of a combo then an edit of that patch would be respected by all consumers of that patch.
It seems that new-comer Futuresonus is taking MPE/ch-per-note seriously with their Parva synth module. I would rather have a DSI synth, but may go with the Parva instead.