It seems like we're going to run into the limits of that by the end of the decade or soon after though. I've made quite a few submissions about this but nobody ever seems to read them :) I really wonder whether we are making advances with parallelism and other technologies fast enough to offset the oncoming barriers to shrinkage and speed.

My naive best guess is that when CPUs stop getting much faster the next wave of innovation will be on bus speeds. But I'm not a chip guy so perhaps my view of the problem is overblown, but I'm very interested in learning more about this from others.