I believe TSMC can produce working 28nm processors right now. That's not in question. What is in question is the yield. This is complicated stuff that they are getting into. It's going to take a while before they work out the yield issues. But they will.
The larger issue is when Medfield out out later this year and it gets compared to TSMC's 28nm processors the question may not become just "which is better" but "Which is better that you can get?"
I think it's going to be similar to 40nm this time with a couple of years worth of fine tuning before achieving optimum yields. Intel will be starting at 14nm at some point when TSMC is still working working the final bugs out on 28nm.
Billions of ARM chips are produced every year, they don't all go into smartphones ;-). Your case is very flawed. If Qualcomm 28nm S4 Krait volume exceeds 32nm Medfield volume in 2012 then the case will be totally dismissed ;-).
The product is still new but 5 of the 8 Newegg matches are "Out of Stock". It is hard to estimate on how full the pipeline is and what the AMD profit is.
AMD shipped 1/3 of their product on 32nm in Q4 just closed. The other 2/3 is on 45nm or larger.
Lots of customers use foundries for their cheapness not their leading edge technology which means they will always be selling high percentages of 'old' process wafers which are fully depreciated in development cost amortization. You are reading far too much into the 10% figure.
10% is no reason to brag. Its quite alarming actually for TSMC & its customers. TSMC CEO has given a lot of "hints" that their 28nm process is in serious trouble, but the longs and ignorant analysts wont hear it until it hits them in the face.