What are the technical challenges to building additional EMPEGish units?

Well, why don't you try it?

Seriously, making general purpose ARM-based computer mobo the size of empeg's is not the problem. Adding top-quality audio-peripherals and making them work well is another story. VFD by itself is OTS, but PIC that drives it and clever interfacing HW/SW combination that makes built-in LCD controller happy with VFD is definitively not. I think that it took some 5-6 HW iterations to bring empeg to where it is now (but now, when the problems are solved and building blocks ready, they can design three similar devices before breakfast).

Kernel is not much of the problem (after all, it is open source, including empeg changes and everything users here did). Flash eprom programming code is also doable, but don't forget that player itself took 10-15 man years to build and perfect. Also remember that in order to use some of programmable components (e.g. DSP) you have to negotiate with manufactuter, sign NDAs etc.

Finaly, manufacture: even seemingly trivial components (like docking sled and its wiring harness) took quite a time to make, cost significantly and are not always as good as empeg guys (and their customers) would like. Even manufacture of another procuction run of empegs (with no design costs) would be prohibitively expensive.

Believe me, empeg's development took years for reason, and it could not cost significantly less than it did before discounts. Of course, if one is willing to compromise at every step, auto MP3 player can be leashed together much more quickly.

But you did say the idea was ridiculous...