A new semi-conductor compound from Intel:

Intel says that replacing silicon with indium antimonide cuts power consumption by ten times while boosting performance by 50 per cent.

New chips employing it are still a decade off, though.

From the article:

The experimental transistors right now rest on a substrate of gallium arsenide, an expensive material used in some communication chips. The company will next try to plant these III-V transistors onto a silicon substrate.

Oh geez. Companies have been trying to do the same thing (put a layer of the III-V material onto Si) with GaAs for years, with meager success. Doing the same with InSb isn't likely to be any easier, I suspect.

Aside from the technical issues (Si forms a very stable oxide that is easy to work with), there has just been too much money invested in the Si chip infrastructure for it to be replaced anytime soon.

And its been this way for a while. Replace InSb with GaAs in that article and you could have had something written 25 years ago.

I did a post-doc at the Naval Research Laboratory in 1985-87. We were working on GaAs and InSb then. I thought this was supposed to be a fast-moving field?

Actually, that could be part of the problem. People were so hot to get something published, that they didn't do an adequate job. I did a series of 15 measurements on a GaAs sample. Everyone who saw the data had the same reaction: "Wow, you really took a lot of data!" Fifteen points? Considering that I was pinning down a curve that sloped up, turned around and sloped down, then flattened out, I thought 5 points per segment was minimal. But then, maybe that was why the curve was still a mystery after more than a decade...

Silicon: present in Earth's crust at an abundance of about 27%.

Indium: present in Earth's crust at an abundance of ~160 parts per billion (by weight)
Antimony: present in Earth's crust at an abundance of ~200 parts per billion (by weight)

It doesn't take a genius to do the math on this one.

Robin: the quantity of material used in chips -- particularly in thin epitaxial layers -- is so small that the cost of the raw material is pretty much irrelevant.