I can't remember where I heard it now, but a while back someone pointed out that every time you shrink in process node size, the cost of development for that 'die shrink' is roughly double the cost of the last die shrink. Thinking it was that Cringely dude...
If you consider 40nm to 28nm a 'full node shrink', this EE Times article would seem to agree with that assessment:
http://www.eetimes.com/document.asp?doc_id=1263318
Here's a .pdf that discusses some of this:
http://www.lithoguru.com/scientist/litho_papers/2013_SPIE_DFM_Keynote.pdf
The above .pdf is slightly dated, and predicted that we'd be down to four foundries by 2018. Well... as of today we seem to be down to just four major foundries: Intel, TSMC, Global Foundries, and Samsung. Here's an interesting article showcasing Intel's node advantage:
http://www.extremetech.com/computing/171477-intels-14nm-milkshake-its-better-than-yours
So, since Intel is worth roughly 2.5 times the value of TSMC and GF combined, it's no wonder why they are roughly 2 nodes ahead. However, assuming that the development costs of those two node shrinks cost roughly 6x (2x + 4x) to accomplish over their competitor's last node shrink, well they need to make up those costs in volume. This development cost disparity is probably (IMHO)the only reason why their competitors are able to hold onto any ground whatsoever.
So it will take a while for GloFo and TSMC to fully deploy 20nm and 14nm (One article noted that TSMC will start deploying 20nm in Feb of 2014, AMD also mentioned 14nm FinFET for low power products sometime in 2014). Intel is already talking about 10nm in that time period.
Samsung is currently flirting with 14nm low power designs as well:
http://www.dailytech.com/Report+Sam...+Exynos+6+Chip+for+Galaxy+S5/article33645.htm
Article about the challenges Intel expects to face in the 10nm, 7nm, 5nm and 3nm transistions:
http://www.extremetech.com/computin...that-will-take-us-to-the-limits-of-moores-law
Quick note: Samsung has somewhere around five times the revenues of Intel ($268 Billion in 2012), but of course they are a much more diversified company (seems like they make everything, from ships to construction equipment to electronics to insurance subisiaries, truly a multinational conglomerate). Nonetheless, They seem to be well positioned to go toe to toe with Intel as they choose, and indeed it is no wonder why they are so strong in the mobile market.
Got a little off topic there, but it's good to understand exactly why AMD and Nvidia aren't on 14nm right now. Kaveri would likely have had a much better wattage to performance ratio if 14nm had been a possibility, but I don't think AMD was ever ahead in the 'node game' to begin with. It's no wonder to me why they spun off Global Foundries - it gave them the TSMC option. But I don't ever see them 'catching up' in the node game.
Again, they need to design smarter if they hope to maintain any semblance of market share. With the innovations they've incorporated into Kaveri, I think they will do OK. And as Toms Hardware and others noted, their GPU business is expected to hit 40% marketshare within the next six months. Yeah, PS4s and XBoxOnes are not PC's, but a GPU is a GPU no matter where you deploy it. Besides, some people don't even own PC's these days (they use tablets, game systems, and their smartphones for all of their computing needs) And AMD NEEDS as many design wins as possible if they hope to stay even somewhat competitive. And, as a few analysts have noted, Intel could use some smartphone wins of their own... and of course they need to stay competitive in the Tablet market to boot!
Kaveri is now the major piece of AMD's 'all in' strategy to secure more of the consumer market, it will be interesting to see if they can hold on to their chunk of the value PC/Laptop market with Kaveri.
And I'll agree, Intel owns the enthusiast market, although some do still flirt with their FXes. I don't see Kaveri putting any sort of dent in that market (it might maintain some enthusiast market share at best, but only in the middle/lower enthusiast segment).
I'm in that middle/lower enthusiast category, so as I noted above I'm excited Kaveri does as well as it does. Waaaaay more powerful than the laptop I'm typing on now...