Hacker News new | past | comments | ask | show | jobs | submit login

That is correct. We would expect some kind of fractional shrink equivalent to the numbering system used.

Intel expects to shrink from 4 to 3 to 20A and finally 18A. Apparently, decent progress has been made on 20A, so at least 3 shrinks should be expected. TSMC has a roadmap to 2nm, and they expect to slow down after that. Samsung has plans for 1.4nm. My guess is that one major player to fall out around the 20, 18, 14 angstrom area. My present thought is that this will be Samsung. I think expect one or two shrinks before Intel or TSMC falls away. One more after that, and then some other technology may need to step up to the plate.

I had long thought that something like gallium arsenide would replace silicon allowing for higher frequencies, but cost might be prohibitive. The world after the shrinks may just be a logic and efficiency war.




I wish we would get a picture with all those metrics.

That said, all of those metrics seem huge to what EUV HighNA can do: I remember near "perfect" 7nm "lines" in photoresist.

I wonder what is the resolution of the motors to align all that (I guess piezo electric). And interferometers for alignment, x-rays??




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: