Weebit - 2022 and beyond, page-1142

8 to 10 months I guess based on previous work. It depends on the shrink and stuff not bumping into each other. If it scales down and all the effects hold between insulators, resistance, leakage and cross-bleed then it will be very straightforward and could easily surprise, particularly if they tested tolerances at 22nm when they did 28nm.

Think of it like a spark plug, gap too small, you get arcing, gap too wide you dont get a consistent spark. Now make the spark plug smaller, do the gaps have to stay the same or can they make them smaller and still get the same effect?

Also, its not serial. Lessons from 130nm, 40nm and 28nm directly applicable to 22nm and they'd have a good understanding of what needs to be altered and what needs to stay the same based on those shrinks. Particularly the move to 28nm from 40nm would have highlighted relevant parameters to pay attention to now.

Again, that said, I dont think 22nm is that important, unless we have a specific customer targeting a SoC at that density and they want uniform process across the whole SoC. 28nm is the sweet spot and we should be prosecuting that as fast as possible. Its the key to discrete and sits under flash density at currently the most economically cost effective density.