I think you're right. Anything that helps TSMC keep key customers away from Global Foundries helps Intel. Unless GloFo can recover development and capacity charges for state of the art performance logic from a larger customer set, Glofo will be forced to provide (and AMD will be forced to use) a generic foundry flow. Those flows are solely based on market demand and the volumes up top are negligable. 4 % of TSMC revenues in 3Q09 were at 40nm ... estimating perhaps 21,000 wafers? As chipguy pointed out you simply don't walk down the yield curve vary quickly on those volumes. You end up with design rules that target volume products where the primary drivers are form factor and lower unit cost, where customers demand first pass yield over a wide range of designs, not performance optimization.
I suspect that is where AMD will end up -- when Intel is converting 22 nm to 16nm -- AMD will be stuck in a 32 nm commodity flow.
Market Data 
Markets 
AI
