Re: - capacity for this year (combination of 130nm + 90nm) approaching 50M, compared to a little more than 30M shipped in 2004, depending on product mix (more server CPUs = less capacity, but higher gross margin parts), with Fab36, depending on die size etc., capacity approaching 100M (with Fab36 fully equipped @ 65nm + fully ramped)
This is a little shocking. It seems like one of Doug's more - optimistic - capacity estimates, rather than anything I would expect. To put that in perspective, if AMD had a 100M unit capacity last year, they'd possibly walk away with a little over 50% market share (assuming they sold all these parts at the expense of Intel's sales, of course).
When is Fab36 "fully equipped @ 65nm + fully ramped"? 2007? If this is AMD's goal, they had better get those commercial client deals won, as well as an account with Dell (better convince them to go 50/50 Intel/AMD, at least), and finally come up with a product line so powerful that it decimates Intel's 65nm dual core generation, including anything based on the Merom core in each of the desktop, mobile, and server segments.
Does anyone besides Doug see this realistically happening? If not, what will AMD do with all this capacity?
And by the way, even the 2005 capacity estimate with Fab30 seems a little high. They are guiding towards a 66% capacity improvement, just from the shrink to 90nm? Seems like a lot of hot air to me, BWDIK...?