material scientist here. I've worked on carbon nanotubes (cnt if you want to sound cool) for a few years. Looking at them under the electron microscope, I've always wondered how in the hell we'd ever line them up without having to recourse to nano-tweezers (which of course is impractical if you want to scale to a few billion switches). Looks like IBM finally cracked a decent technique (though I suppose the device yield isn't that great), and it hopefully will be a path toward production. What would be really cool is if we can find economically-viable application for carbon nanotubes outside of semi-conductors. The biggest problem right now is their cost (~$400/gram for good purity [for reference: 20X the price of gold]) and the fact that platinum is used as a destructive catalyst to make them.
Ah, good catch! I remember a while back comparing CNT prices per gram versus gold and it was something like $20 a gram for gold. Must have been a low then. Thanks for catching it.
Well, right now their yield is only 70% of the transistors, so effectively 0 for chips. Improving that is probably the biggest research breakthrough still remaining.