What happens to an amp below 2 ohms?

Very few amps will handle such a load but I know of only 2 speakers myself that go below one ohm: the aforementioned Scintilla and the Quad 57. The latter does this in the top end so is not nearly as difficult a load to drive as the Apogee. In a perfect world the impedence curve would stay above 6 ohms but in our world some very good speakers have difficult curves. My Gamut L5s go down to 3.2 or so and my friends Sasha to below 3. The amount of work my CJ 350 has to do driving the Gamuts is considerable, gets quite hot driving them to a good volume in a long session. On the other hand it gets barely warm driving my Spendor S 100s to a similar volume. A good reason why you should always buy the speaker first; then you know what your amp requirements will be.

The Krell KSA and MDA are noted for playing the Apogee line of speakers with aplomb. In fact Krell put themselves on the map by building amps that could do this and were the amps of choice for Apogee to demo with. I'm not sure if some of the latter Krells could still do this but I'd be surprised if they couldn't.

The benefits that some speakers that have low impedance's provide, outweigh the disadvantages of that low impedance. Buy the speakers you like best, then buy an appropriate amp.

They blow up and make a mess!

Buy amp and speakers as a unit...and than after a nice test listen. If you are really after a 2 ohm stable amp, and they are few/farbetween, leave out ALL the 'd' amps with ICE modules. The data sheet says the limit is 2.7 ohms. I'd be inclined to listen to 'em.

I'm not even gonna touch the 'need' for a 2 ohm speaker. What I will say, is that if it were pure resistive, you'd have better luck. Since it isn't, you have to deal with reactance. bad juju. You may get lucky and it won't be too awful, but in more extreme cases of large phse angles (highly capactivie or inductive) you're gonna suck up amp power like you own stock in the power company.

Hi sensitivity would also help, but only so much. Less phase shift would buy you more than a sheerly high sensitivity number. especially if you knew such phase shift peaked at or near the impedance minima.

In general, however, there's no need for such a wacky load.