AMD's push for 30% marketshare begins today

[YKhan]

AMD's 300mm Fab 36 opens today, 14-Oct-2005, they expect eventually it
will account for 30% of global PC processor sales.

With new factory, AMD ups ante against Intel | Tech News on ZDNet
http://news.zdnet.com/2100-9584_22-5895124.html

 

[keith]

AMD's 300mm Fab 36 opens today, 14-Oct-2005, they expect eventually it
will account for 30% of global PC processor sales.

So even if they succeed (go AMD , they're closing Fab 30 tomorrow?

 

[YKhan]

I think they're keeping Fab 30 also. They'll probably wait until Fab 36
is mature enough, and then they might retrofit Fab 30 with 300mm
equipment too.

Yousuf Khan

 

[keith]

I think they're keeping Fab 30 also. They'll probably wait until Fab 36
is mature enough, and then they might retrofit Fab 30 with 300mm
equipment too.

I was making a tongue-in-cheel comment about the phrase "it [referring to
Fab 36] will account for 30% of global PC processor sales". Unless
they're planning on 50% market share, I infer than fab 30 is going away.
Not that I'd mind AMD with a 50% market share. ;-)

 

[YKhan]

Actually Hector Ruiz was saying that by the time Fab 36 approaches 100
million processors a year, at that time he expects that amount to be
only worth 25-30% of the overall market. So he's assuming the market
will be between 300 million and 400 million processors by that point.
Fab 30's 50 million processors is going to look like small potatoes by
then, and it's likely it will need to be retrofitted to 300mm or start
producing chipsets.

But I do think 300-400 million processors a year as the overall market
is extremely optimistic. It would depend on Intel growing heavily too.


Yousuf Khan

 

[George Macdonald]

AMD's 300mm Fab 36 opens today, 14-Oct-2005, they expect eventually it
will account for 30% of global PC processor sales.

With new factory, AMD ups ante against Intel | Tech News on ZDNet
http://news.zdnet.com/2100-9584_22-5895124.html

Funny but based on reports I'd been reading, including this article, I had
been expecting Fab 36 to be 65nm from the get go but apparently not... all
other articles, including the presss release at AMD say it'll be 90nm and
they don't expect *any* 65nm production until end 2006 and full conversion
in mid-2007.

I suppose Intel is the one with the pressure on it to come up with
performance-wattage-area balance and AMD's 90nm seems to be working
extremely well, at least from my personal experience -- my 3500+ Winchester
core barely cracks 50C at full tilt -- but all the same, I hope AMD is not
opening the door just a crack here. Intel could make a lot of progress
with their process in 18months.

 

[Yousuf Khan]

Funny but based on reports I'd been reading, including this article, I had
been expecting Fab 36 to be 65nm from the get go but apparently not... all
other articles, including the presss release at AMD say it'll be 90nm and
they don't expect *any* 65nm production until end 2006 and full conversion
in mid-2007.

Yeah, that's what I've been led to believe as well until now too.

I suppose Intel is the one with the pressure on it to come up with
performance-wattage-area balance and AMD's 90nm seems to be working
extremely well, at least from my personal experience -- my 3500+ Winchester
core barely cracks 50C at full tilt -- but all the same, I hope AMD is not
opening the door just a crack here. Intel could make a lot of progress
with their process in 18months.

Well, at 90nm that would mean that AMD will have both Fab 30 and 36 at
the same process which would be the equivalent of 3 times their existing
Fab 30-only capacity (assuming that 300mm @90nm is twice the number of
dice as 200mm @90nm). Then there's also Chartered outsourcing for them.

Yousuf Khan

 

[George Macdonald]

Yeah, that's what I've been led to believe as well until now too.


Well, at 90nm that would mean that AMD will have both Fab 30 and 36 at
the same process which would be the equivalent of 3 times their existing
Fab 30-only capacity (assuming that 300mm @90nm is twice the number of
dice as 200mm @90nm). Then there's also Chartered outsourcing for them.

Oh sure they'll have the increased volume from wafer size but Intel will
have crowing rights for the spin-doctors and anal...ysts. I'm a little on
the fence, but hopeful, about Chartered - they have been shakey for about a
year now and we'll have to see if they can pull it off... with maybe a
little help from IBM?

I'd also thought that DSL had been present in AMD's chips since early this
year; in fact I'd thought that was how they managed to ramp up the clocks
on the 90nm process after the limited clock Winchesters but according to
this article
http://www.technologyreview.com/articles/05/02/wo/wo_brown020305.asp?p=1
admittedly from back in Feb05, it should be coming out about now... so
something to keep an eye on. Could be that AMD has decided they have
enough in the bag to stay with 90nm for another year... if true, something
they could do a better job of with leaked PR to the usual investor house
monkeys. I mean all this crap about a cobbled together dual "solution",
65nm, leapfrogging etc, and some fundamental process advance by AMD/IBM
gets "missed"??

Further thought on strained silicon: I see that, whereas AMD has already
paid the piper on that, Intel is being sued by Amberwave for patent
infringement... interesting.Ô_ô Dunno where IBM stands here but one of the
original researchers, Jeffrey Welser, is now an IBM employee... interesting
background here http://www.sciencenews.org/articles/20040228/bob8.asp.

 

[YKhan]

Oh sure they'll have the increased volume from wafer size but Intel will
have crowing rights for the spin-doctors and anal...ysts. I'm a little on
the fence, but hopeful, about Chartered - they have been shakey for abouta
year now and we'll have to see if they can pull it off... with maybe a
little help from IBM?

Well, Chartered is in it for 90nm only, so far. From what I understand,
even after AMD switches to 65nm in its own fabs, Chartered will produce
90nm versions. Though Chartered said they are capable of producing at
65nm too, I don't think AMD wants to take the chance on that yet, with
Chartered. It might let Chartered go 65nm, once AMD has gotten its own
65nm process sorted out. I assume that Chartered's output will require
separate OEM validation from AMD's output.

But then again, I was led to understand that AMD was going to start off
with 65nm on Fab 36, and here they are starting off at 90nm.

I'd also thought that DSL had been present in AMD's chips since early this
year; in fact I'd thought that was how they managed to ramp up the clocks
on the 90nm process after the limited clock Winchesters but according to
this article
http://www.technologyreview.com/articles/05/02/wo/wo_brown020305.asp?p=1
admittedly from back in Feb05, it should be coming out about now... so
something to keep an eye on. Could be that AMD has decided they have
enough in the bag to stay with 90nm for another year... if true, something
they could do a better job of with leaked PR to the usual investor house
monkeys. I mean all this crap about a cobbled together dual "solution",
65nm, leapfrogging etc, and some fundamental process advance by AMD/IBM
gets "missed"??

I haven't heard much about Dual-Stress Liner either recently. Don't
know if it's already in place or not.

Further thought on strained silicon: I see that, whereas AMD has already
paid the piper on that, Intel is being sued by Amberwave for patent
infringement... interesting.Ô_ô Dunno where IBM stands here but one of the
original researchers, Jeffrey Welser, is now an IBM employee... interesting
background here http://www.sciencenews.org/articles/20040228/bob8.asp.

Hmm, Amberwave's patents were about strained silicon? I had assumed it
was about clockspeeds or something. Too many patent lawsuits to keep it
all straight I guess.

Yousuf Khan