CVTs have a reputation for inefficiency and for slippage. That's not a big problem in industrial equipment (I don't care of my 3hp Bridgeport milling machine motor delivers 2.8hp to the spindle due to losses in the cone-pulley CVT, after all) but it's obviously a big deal in a bicycle. With the balls, rings, and fluid, I also have to wonder how heavy it is.
It's a neat device, and if it really works as well as they imply it could be pretty cool. I'll wait until I see hard numbers for efficiency before I get too excited about it, though; the trade magazines are full of innovative, breakthrough mechanical products that somehow seem to consistently fail to live up to their advertising hype.