> Each of the last productivity multipliers coincided with greatly expanded markets

They didn't. But it may be a relevant point that all of that was slow enough to spread that we can't clearly separate them.

Anyway, the idea that any one of those large markets is at saturation point requires some data. AFAIK, anything from mainframe software to phones has (relatively) exploded in popularity every time somebody made them cheaper, so that is a claim that all of those just changed (too recently to measure), without any large thing to correlate them.

> That's quite unlike previous revolutions where we had to build all the software from scratch

We have rewritten everything from scratch exactly once since high-level languages were created in the 70s.