Fabrication is a function of cost. If your chip's performance is good enough at a lower res node, simple cost benefit analysis clearly indicates that going to the next process node and adding the cost of fabricating finfets etc etc is simply not worth it for most chip applications. For Intel which is the leading light for high performance consumer computing it makes sense to make the huge investments necessary for further extending Moore's law for PCs, but for the average smartphone vendor, paying more for higher performance chips has become a case of diminishing returns. In other words while it maybe worthwhile to invest in more complex fabrication in the PC space specially now when it is expected Intel and Microsoft will completely change the landscape by integrating tablets into the fold, not in the smartphone space. The only reason all these foundries are investing in fab upgrades is because they think Intel might have an advantage with fabrication in the smartphone chip space. Over time they will realize that this is not the case. They're just playing it safe. The csenario that will probably play out for the foundries is that their most popular process node remains at 28nm even though higher res nodes are available.
khitchdee, you #$%$ did you even read the post? It talks about how Apple screwed the pooch by not having a superior node to move to. Just like that node wasn't enough, 28nm isn't going to be enough. Why? Because new functions are constantly being added and without improvements in fabrication the chips get too big to be practical. Apple got a serious lesson in this. What's your problem???
The only reason all these foundries are investing in fab upgrades is because they think Intel might have an advantage with fabrication in the smartphone chip space. Over time they will realize that this is not the case.
Yeah right - 9 months ago Apple launched a 45nm I Pad and it tanked.
This is one of the stupidest post I ever red...
why don't you check TSMC projected capex dummkopf
Some people, the early adopters who skim off the top end of the market. The volumes are somewhere in the middle. If you're trying to create a high volume product, the market is not very feature sensistive and it is a little cost sensisitive so you cant justify a higher cost higher performance chip that is adding a few features over the lower cost mainstream. It's like spuer-computers and PCs.