Why did Nikon choose 36 megapixels ? - why not larger ?

Unlike most chips, sensors with smaller features cost more to make because their overall size doesn't go down. Whereas with each shrink Intel get more chips off a wafer, Nikon and Sony don't. In fact, yields go down because as the number of photosites increases, the number of errors does.

Current cellphone chips scale up to full-frame with 100 MP or more. We will get to these numbers but manufacturing will have to improve in order to keep prices reasonable.

I wonder if the solution will be 200 MP+ sensors with bin pixels and output lower resolutions allowing for individual photosites to fail without damaging the output.