That's not the basis of my argument. But consider it "backgrounder information" that I would think would serve as a big flashing "Caution" sign to those who venture out on a limb like you did.
Your claim constrained NOTHING about the processes used, now, or incredibly, in the *future*, merely that they agree on feature size and supplied voltage.
Even attempting to repair your claim, inserting additional necessary constraints, but not unrealistic ones, it seems highly unlikely you can conclude anything useful. You aren't going to get power-per-computation within an order of magnitude without artificially quashing all innovation, and ignoring existing data, so why make the attempt?
To use your analogy, we're not talking carbon-nanotubule bridges, but different companies may temper the steel differently, or experiement with composite materials in the decking, or even invent something we haven't thought of yet. Imagine that.
Face it, the whole exercise was an attempt to say, "No one will be able to do it better than Intel did in Prescott," and you tried to back that up with a big pile of pseudoscience.
Doug
Market Data 
Markets 
AI
