OK, let's look at this claim carefully. Atoms are around 0.5 nm in diameter with silicon being about 0.25 nm. I'm sorry but I just can't picture an 8 nm pipe made out of silicon which would span just 32 atomic diameters. The electrons would be the size of boulders rolling down a creek! Not to mention how you're going to keep them from Heisenberging themselves where they don't belong. Is this Intel dude talking out of his hat or is this even remotely possible?
At that size it is unlikely to be very similar to the CMOS process used today, likely not even silicon based. 8nm is possible, theoretically, its just going to take a while to do, but 7 years is quite a ways away and gives them quite a while to figure it out. If in the 50's you had told some one that we would be putting billions of transistors on chips smaller than postage stamps, and they would be only 45 or 32nm longs, they would have told you it was impossible. For the semi conductor industry, 7 years gives them quite a bit of time to get the hang of things, i find 8-10 years a bit more plausible, but 8nm can be done, in theory, we just need to keep working on doing it in practice.
I agree that its doubtful, but then again we should be at 32nm for much of 2011, and 22nm for Intel isn't far behind that.
Then again it could be another claim like they said during the GHz race. Remember that the P4 was supposed to hit 10GHz at some point? We'll have to wait and see if they get to 8nm, but like the GHz claim it doesn't really mean much. We will continue to get faster and faster CPUs, the tech behind it isn't THAT important.