AMD to abandon SOI after 45nm

[Idontcare]

After 45nm, there won?t be any SOI process and 32nm for both IBM and AMD. This will mean the return to Bulk process with High-K in mind.

http://www.fudzilla.com/index....=view&id=9869&Itemid=1

So much for the the ZRAM rumors that make their rounds every now and then.

 

[Martimus]

Originally posted by: Idontcare

After 45nm, there won?t be any SOI process and 32nm for both IBM and AMD. This will mean the return to Bulk process with High-K in mind.

http://www.fudzilla.com/index....=view&id=9869&Itemid=1

So much for the the ZRAM rumors that make their rounds every now and then.

I thought IBM was planning this move at 22nm. I never believe anything from Fudzilla anyway, so maybe they still are (or I have things mixed up in my head - which is more likely.)

 

[Idontcare]

It'll only be confirmed when AMD denies it.

It makes technical sense (to abandon SOI post 45nm), and that may be the sole basis of Fudo going out on a limb and fabricating a story about it being AMD's future.

However it does put AMD (Foundry Co) back onto the "me too" treadmill where they lose yet another differentiating feature to their process flow.

If you want a "me too" process flow with about a 12-18month lag time behind Intel then you don't need the Foundry Co for that, you had TSMC as an option there all along.

Personally I tend to agree with just about every other industry professional published opinion out there in that the foundry market was already saturated and in need of consolidation before AMD announced this spin-off.

I won't be surprised at all if Foundry Co doesn't exist when the ROW is rolling 22nm production. They'll merge with UMC or Chartered in my opinion within the next 24 months.

 

[hans007]

Originally posted by: Idontcare
It'll only be confirmed when AMD denies it.

It makes technical sense (to abandon SOI post 45nm), and that may be the sole basis of Fudo going out on a limb and fabricating a story about it being AMD's future.

However it does put AMD (Foundry Co) back onto the "me too" treadmill where they lose yet another differentiating feature to their process flow.

If you want a "me too" process flow with about a 12-18month lag time behind Intel then you don't need the Foundry Co for that, you had TSMC as an option there all along.

Personally I tend to agree with just about every other industry professional published opinion out there in that the foundry market was already saturated and in need of consolidation before AMD announced this spin-off.

I won't be surprised at all if Foundry Co doesn't exist when the ROW is rolling 22nm production. They'll merge with UMC or Chartered in my opinion within the next 24 months.

well i don't think that foundry company really afffects the independant foundry business that much as much as being over saturated.

i mean, in that case, the foundry company is adding capacity to global capacity, but AMD not having fabs, adds to global demand, so assuming AMD continues to use all of the foundry company's capacity it won't do anything really to the over or undersaturation of global fabs.

 

[Idontcare]

Originally posted by: hans007
i mean, in that case, the foundry company is adding capacity to global capacity, but AMD not having fabs, adds to global demand, so assuming AMD continues to use all of the foundry company's capacity it won't do anything really to the over or undersaturation of global fabs.

AMD's fab capacity consumption is not a NY fab (yet to be built) nor the mostly untooled fab in Dresden, nor the well underutilized fab in Dresden.

Presumably the Foundry Co is going to want to run wafers thru all three fabs at around 85-90% capacity. That requires them to more than double the current wafer starts that AMD generates.

This requires a heck of a lot of customer demand for wafer starts and that demand must be won away from TSMC, UMC and Chartered.

Were the foundry Co not interested in tooling out the Dresden fab or building the NY fab then I'd agree it doesn't matter whether AMD owns the fabs or the Foundry Co as the supply/demand for wafer starts could remain unchanged in that situation.

But presumably the Foundry Co intends to increase their volume of wafer starts and that requires a LOT more business than just AMD's offerings.

I've personally seen a fair amount of how the foundry/fabless/fab-owning industries work, and there is a shockingly narrow window of opportunity here for the Foundry Co to successfully compete with and survive against the TSMC machine. (consider that fact IBM's efforts in this identical pursuit have failed miserably as well)

The Foundry Co stands to offer really high performance CMOS technology perhaps 12-18months before TSMC gets around to it...depending on when TSMC actually qual's their 28nm HK/MG that they claim will be released in 2010.

If they don't beat TSMC to that milestone, and generate confidence with the fabless world that the Foundry Co is a company the fabless guys can risk their own future business model with, then it will be game over. Not too unlike SMIC.

I wish I could be far more rosey about the Foundry Co's prospects, but I've got too much data in my head about how this part of the industry works (and what doesn't work) to just ignore it all and think this is going to be an awesome way for the underdog to win again.

Now AMD can certainly do well here. If the Foundry Co dies it will be because it has competition (TSMC) and that means AMD itself will have options and alternatives to the Foundry Co which means they stand good chance to continue to live.

 

[bryanW1995]

you can rest assured that two things will happen: the foundry company will do poorly, and hector ruiz will get a very large bonus when it, um, "founders".