[Here is another dose of reality. Note that the comments apply to ARM fabrication only...]
28nm is the first process shrink which doesn’t deliver a cheaper chip, says Scott McGregor, CEO of Broadcom.
According to McGregor, not only is 28nm more expensive than 40nm now, as might be expected at the start of a node, but that, on Broadcom’s current projections, 28nm will never be cheaper than 40nm across the whole lifetime of the node.
“What we're seeing in the industry is the cost of next-generation nodes rising exponentially," says McGregor, “what this means is, unless you need the advanced process because of performance reasons or die-size reasons, you're not going to get a cost benefit from converting to the new node."
McGregor went on to say that the same would be true at 20nm.
The consequences will be dramatic: companies will stick longer with a node, foundries won’t be so aggressive in their process development and the old learning curve of chips getting cheaper by 30% a year will decelerate.
Apple, reportedly, has stuck with 40nm and 65nm technology rather than risk its products on the availability of 28nm parts.
ARM isn't in the fabrication war with Intel... TSMC and IBM are, to name two. In the logic design war, where ARM and Intel do both battle, the winners and losers are less clear. Smaller geometries can lead to lower power, faster clock frequencies, greater density, and higher overall performance, but don't discount innovation and leadership in basic architecture and logic design.
IMHO, INTC is priced the way it is because the real battle is in chip architecture, not in fabrication processes, and INTC isn't the clear winner in architecture beyond PCs and servers yet. It's not so many years ago that AMD gave Intel a black eye in microprocessor architecture that took Intel a while to recover from. Yes, fabrication superiority played a role in Intel's recapture of clear leadership in laptops and servers, but it wasn't nearly the only thing.
What this says to me is, "28nm is the first node that isn't cheaper, UNLESS YOU ARE INTEL."
The same thing happened at 120nm going to 90nm, leakage went up, performance went down. Intel addressed that with HKMG...(who ever heard of a use for Hafnium!??). Of course 28nm from TSMC will cost more because they can only make a little bit of it. It is pointless for anyone but Intel to move to 22nm without Trigate, because the same thing will happen as happened at 120nm. 22nm planer processes will produce garbage products.
I seriously think that 28nm is the end of the line for the foundries. The amount of money that they would have to spend to stay one or more generations behind Intel is prohibitive.
Dan Whatever can blow all the smoke he likes about 28nm being healthy at TSMC. In the unlikely case where he is right....we're talking about 28nm planer....who cares? If their 28nm was running perfect they are still a node behind...two nodes if you consider Trigate a node.
None of this means that foundries and the fabless model goes away. If Intel doubles their market share to ~30%, it means that $.70 of every semiconductor dollar will be non-Intel. TSMC can easily carve out $20 bill of a $300 bill market with 28nm and above processes.
Big chips that can benefit from 22nm Trigate and below, whether it is cost or performance, will find their way to Intel.
To those out there that don't understand this crazy business we might as well be speaking Martian, nobody understands how dominant this makes Intel at the bleeding edge of technology.