Interesting piece about how silicon fabs ar getting more and more expensive and so we might see Moore’s Law slow down.
I’m not sure he’s entirely right in detail, I think there might be some confusion between cramming the transistors on closer together and using bigger disks to make the chips from. From what I understand it’s that latter that dramatically raises the costs.
Still, I think there’s one huge thing missing from his piece:
Moore’s law has seen computing power grow exponentially for 40 years – but soon economics, not physics, may stop it.
He’s using the word “economics” there to mean huge costs. But buried in there is the real argument, that the huge costs are indeed there but no one is sure that they can sell enough of the new chips to make those costs worthwhile. That is, the proper economic point being that the costs will be greater than the benefits (or the benefits the consumers are willing to pay for perhaps) and thus this is an investment that we don’t in fact want to make. That is, economisc is stopping us from wasting scarce capital.