bobs10, one thing that you should add to your calculation is that by the time Fab 36 comes up, Intel will have fixed their current mistakes. Intel will spend 2005 productively developing new desktop & Xeon processors based on Dothan derivatives with iAMD64 and multicore added. Expect them to be very competitive.
AMD needs to build a larger market presence before Fab 36 comes on line. AMD's product advantage is not forever, that is why they are looking into outsourcing arrangements now. They know that they had better build solid OEM relationships and increase marketshare before Intel catches up, or they will have an overcapacity problem in 2006.
While the concept of A64 fabbing at IBM is intriguing, I think that Sempron (K7 and/or Geode) outsourcing is more likely. This frees up Fab 30 to focus on volume K8 while building overseas marketshare with low cost product.
Although I agree with Ruiz' priorities of raising ASPs and showing value for the product, I also know that raising marketshare has to be the followup priority if AMD is to break out of being a marginal x86 manufacturer. Thus a sound strategy is to have Fab 30 continue to maintain 15% marketshare, but on K8, while the low end is fabbed externally to achieve another 15%.
Just because AMD builds it does not mean "they will come"!
AMD has a window of product opportunity which closes at the end of 2005, they must take advantage of that opportunity now. 2006 would be too late.
News 
Market Data 
Discover 
