facebook rss twitter

AMD ramp up capacity to meet Intel onslaught

by Ryszard Sommefeldt on 31 May 2006, 09:07

Quick Link: HEXUS.net/qafuf

Add to My Vault: x

Monday saw a somewhat quiet press release announcement from AMD, letting the world know that Fab30, their famous processor fab in Dresden, Germany, is to get a name change and a major upgrade. With Fab36 already processing processors on 300mm wafers for AMD on the same site, Fab30 has been a 200mm workhorse for some time and, faced with increasing demand for their popular CPU products, an upgrade to 300mm was almost inevitable we suspect.

Making chips at 300mm means more than double the number of working dies per wafer, assuming the same small defect rate, which will allow AMD to help meet demand. Combine that with a transition to a 65nm process node on both their 300mm fabs by the end of next year, and assistance in manufacturing from Chartered Semiconductor, and this $2.5B investment looks well spent. The major upgrade of 30 to 300nm will see it get a name change to Fab38.

AMD are also looking ahead to 45nm wafer production in an agressive timeframe, letting their CPUs get bigger in terms of transistor count for the same overall die space and wafer use, or even smaller with more coming off the boats.

With a stern upcoming challenge to meet from Intel and Core 2 Duo and expectations high on their next generation consumer CPU core -- codenamed K8L -- the transition to larger wafers at 65nm will be a key one for AMD's continued success. The CPU world gets ever more interesting by the day, so keep an eye out to see what both major vendors get up to in the coming months.

Will Fab36, Fab38 and Chartered's capacity be enough on the production side? We'll watch closely and see.

HEXUS.links

Full press release
AMD's Socket AM2 transition and FX-62 performance evaluation
HEXUS' exclusive preview look at upcoming Core 2 Duo E6xxx series CPUs versus FX-62 and FX-60


HEXUS Forums :: 5 Comments

Login with Forum Account

Don't have an account? Register today!
If a company like Dell come along and say they want to buy your product in volume, do you say “hmm, sorry I don't think we can produce that much” or do you make damn sure you can?

That said though, I wouldn't be surprised if there was a psychological edge to an announcement like this. If I were an Intel exec (even with Conroe on the way) and I read this news, I'd be thinking “these boys aren't messing about here, that's a big spend so they must be sure of worthwhile returns”…
I don't think thats the reason at all - AMD have had to do this for years. They won't make money in the margins which Dell work to but it does mean they can ramp volume.

Currently AMD don't do all of their production; some of it is outsourced.

This is big boy money to spend - you don't do this just for a deal with Dell - you do this because you have to.
well, this can only be a good thing, means cheaper processors eventually!
Deleted
well, this can only be a good thing, means cheaper processors eventually!

Once they've got that $2.5bn back :p
Deleted
I don't think thats the reason at all - AMD have had to do this for years. They won't make money in the margins which Dell work to but it does mean they can ramp volume.

Currently AMD don't do all of their production; some of it is outsourced.

This is big boy money to spend - you don't do this just for a deal with Dell - you do this because you have to.

I didn't really mean in terms of fiscal returns (from the Dell contract), more that being such a significant win they're not going to want to upset Dell by not being able to produce. I imagine the Dell Opteron offerings are going to be in high demand and similarly Dell are going to want to expand on the range of Opteron machines as well. It seemed logical to me (based on my assumptions) that the Dell thing might be at least one factor in AMD's fab capacity expansion.

But you're right enough, if AMD are spending this sort of cash they must be damn sure of ROI, how much of it is from Dell will only be a part of that. Outsourcing anything is only really worth the savings up to a point; beyond that point it becomes more cost-effective to bring whatever it is in-house.