If 28nm turns out to be a "resting place", then 28nm FDSOI (w/perf & power equivalent to 20nm planar bulk but at a 28nm price), should certainly be a contender. Love takes time, @RickMerritt, and FinFETs had a lot more time (&$$) to get traction. But a better deal is a better deal, so don't write FDSOI off yet! Also, Leti will have the 10nm FDSOI PDK in June (yes, in just a couple months), so not sure why people are saying that FDSOI won't go the distance (my understanding is that it will go up against 7nm bulk FinFET -- much like 28nm FDSOI now goes up against 20nm bulk planar). As many have pointed out, competition is a good thing -- so really, both FinFET and FDSOI should coexist, serving the markets to which they're best suited, right?
Strictly speaking, 28 nm node is actually something like 45 nm half pitch (minimum), which is also very difficult and design-limiting for single-patterning lithography, being very near the resolution limit. It may therefore not necessarily give as effective a shrink, and possibly long lines still need to be cut, which may still be considered pseudo-double patterning. So possibly 40 nm is suitable as a "resting" node as well. You could have SMIC "resting" at 40, TSMC at 28, and Intel at 22. Maybe Samsung would still charge to 14?
@resistion, i think your argument here is totally valid, giving us some interesting perspectives. After all, we tend to focus on the latest node all the time (the nature of the media biz), but in reality, there are a lot of chips to be made in the current and previous nodes.
It has been the general impression that Intel 14 nm would arrive soon while SMIC 28 nm would always be delayed. The surprise is the other way around. It would disturb ("panic") those who have held the former stereotype.
It's not to criticize Intel or praise SMIC. SMIC has no reason to rush 28 nm, but it can only gain by getting there. Intel has lots to lose to not get to 14 nm, but it should be the first to get there; clearly it's more difficult than previous nodes.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...