I mean, the issue with his assumptions (that are really not his assumptios but a well known vodoo god-like belief of blind technologists) is that theg do not take into account tail events. Like a global war, for instance. Or the invasion of a country that owns the 80/90% of the esportations in chips (as the Taiwan/Chinese drama).

These events terribly change the forecasts outcomes but we cannot really model the probability these events.

Thus, population growth is still a preferrable situation.