Hi Gattomanzo.
Despite Next Limit's behavour, and uncertainties over how well their support may or may not be, I do think it is worth including Maxwell as part of your training program, if only to show another aspect of the rendering possibilities.
However, I would be a little reluctant to base your entire rendering lessons on Maxwell. Two reasons, first it is untested in the wider industry, and second I would be tempted to teach software that your students are more likely to use in the working environment.
At present I'm not sure how well Maxwell will fair, even in industries most suited to it's strengths, i.e. architecture.
Whilst it's true that more than any other clients, architects appreciate good quality lighting, and will be prepared to pay a little more for better lighting simulation, even architects wont pay too much over the odds for overlong render times.
I generally present new clients a range of render samples (from flat shaded, to photorealistic) together with likely pricings for each. Whilst just about all really like the Maxwell renders (mainly for the natural external light quality), I've not had one that was prepared to pay the asking price. Most usually opt for the almost as good, but a lot cheeper Vray renderings.
I hear all the time people on the Next Limit forum (generally from members of the A-Team), that Maxwell whilst relatively slow at actual rendering, they find they are as fast as other renderers, if not faster, because they can predict how the materials will react and so enter material settings right at the start without having to do loads of test renders. Personally I find this puzzelling, because if you have used any renderer for any length of time you have a fealling for how materials will react right from the start.