Otellini Hints Why In The End, There Will Be Only One

Otellini: "Chip manufacturing gets harder and more expensive all the time. It's basically the laws of physics and the laws of economics at play. To build a modern semiconductor plant which uses 300-millimeter wafers, and state-of-the-art lithography with designs measuring 45 nanometers, costs close to $4 billion. Then there are the tools inside it, which amount to close to a billion dollars, per generation of technology. That means you need to generate $4 billion to $5 billion a year in revenue out of it. There are not many semiconductor companies that are $5 billion or more in revenue. So economics leads many companies to collaborate. "

AMD currently generates roughly $5B annual revenue but runs 2 Fabs. It is clear that the reason why AMD needed to capture 30% marketshare was to generate enough revenue to support 2 Fabs. Having failed that goal, it will be tough for AMD to return to profitability without significant sacrifices. We've seen a few adjustments, one of which was to halt CAPEX on 300mm upgrades for Fab30. This will affect the company's long-term competitiveness in cost and still this isn't enough to turn the company around.

Otellini: "No. 2: The laws of physics mean chip making gets harder and harder. We think we have some breakthrough technology at 45 nm. We think it will be harder for people to do that than it was in the past at, say, 65 nm or 90 nm. Our lead over the competition may extend with this generation and probably extend a lot after that. This is not a macho thing. It's all based on sheer economics. "

AMD's reliance on Foundries to support volume needs at the low end is not a viable long term solution to compete with Intel. The diverse set of customers Foundries have forces it to focus development on process shrink and less on transistor performance for logic devices. As Intel rolls over its volume runners to 45nm, AMD's foundry made processors will look 1-2 generations behind. These single core 90nm, foundry made CPUs will never command a high enough price to make any business sense. AMD can go on and continue the price war with useless processors but AMD needs to understand that the quicker they commoditize CPUs, the quicker their demise will be. We've already seen what the lastest price war has done to AMD's profits - or lack of.

A Talk With Paul Otellini.


Anonymous said...

Bye bye AMD.

Anonymous said...

Good riddance to AMD. They've been nothing but a plight to everyone. Trying to ripoff consumers with $350 X2 3800s. Just the thought of AMD disgusts me. Once AMD is BK the world will be a better place. I wholeheartedly thank Paul Otellini for eliminating the AMD plague.

Anonymous said...

I feel really bad for AMD employees who are being kept in the dark about Barcelona. Only a few employees really know the real story which is that there will be no chance in hell that Barcelona, the quad core version, the one that is supposed to have high margins for the company, will be DOA because of what Harpertown will do. This is why you see AMD rhetoric in Barcelona beating Clovertown but you don't hear squat about Harpertown which is the 45nm version. AMD are trying their damnest to not be shot at Computex: the rumor is that Intel will let the general press bench Harpertown at Computex whereas AMD will not take that approach.

I am sad for the hard working employees at AMD because they are the ones tht are going to ultimately suffer the same fate that the Intel employees went through last year. when Intel chopped it's 10,000 people it did so with a winning product. Here the situation is different for AMD.

Heat said...

Hahahhaha dont tell this to scientia and sharikou. They still believe that AMD is speeding up 45nm and will release it close to when intel releases their 45nm processors. AMD has never been close to Intel on process technology and they havent even gotten their 65nm close to intel's left alone already migrating to 45nm.

Anonymous said...

We are a month away from the AMD Q2 earnings announcement and unfortunately within a couple of weeks of an AMD Q2 earnings shortfall pre-announcement.

At least AMD was wise enough not to introduce a new socket with Barcelona. At least two OEMs have now been briefed on Barcelona rollout schedules. Hector was right: no material impact this year because they know they wil lnot be competitive.

The only thing that may save AMD is socket F and 3GHz Opteron K8 parts because that is the only thing they're going to have to

Anonymous said...

You are off by a month.

It woudl be a miracle for Intel if AMD preannounced anything this month LOL

Anonymous said...

AMD Phenom FX/ X4 in late Q3 early Q4


Not to be seen at your local mom and pop or retailer near you this holiday.

AMD BK 2Q'08

Reverse 10-1 AMD stock split upcoming.

Anonymous said...

Interesting read but I think you are a bit pessimistic with the #'s, especially when you state to "support two fabs"...

As Otellini mentioned once the fab is built you need ~$1Bil capex for every gen (2 years) plus obvious operating cost/materials.

So maintaining 2 fabs and upgrading them on a 2 year cycle is about $1Bil/year in capital spending. Thus once you get past the initial outlays (which of course is no minor feat), it is sustainable with reasonable revenue rates. If they were to follow Intel's model and skip a generation with each of the fabs (90nm-45 in one fab, 65-32nm in another, 45nm-22nm, etc...) that also may help them a bit. Then you are talking about upgrading each fab every 4 years (though the amount of capital needed would go up as you would have less reuse). You also have less capacity disruption as you don't have constant upgrade cycles while trying to produce revenue...

The real issue is operating costs, materials and the R&D to sustain that 2 year upgrade cycle is becoming ever increasingly expensive.

So what AMD needs (which is not simple by any means) is the ability to ride things out until the upgrade of F30 to fab38 - that is the real key to their long term viability. If/when both fabs are up and running then you are talking about $1Bil capex to sustain both fabs which is I think achievable (even with less than 30% market share).

The real question will be how much is needed for R&D - yes they have an agreement with IBM, but I don't think IBM has the same sense of urgency on node scaling as AMD now does. This to me is where Intel's key advantage lies in the future.

Another wildcard is whether or not Intel pursues 450mm. Right now only Intel and Samsung can really afford to do this (maybe TSMC). Intel has been pushing ~2012 for this and while this is still a bit out, if they managed to do this (though I doubt will happen), it would give them a SUBSTANTIAL cost advantage. While this is far out in the future the R&D spending and planning would start picking up in 2009/2010. This would force AMD to a tough decision on whether or not to go this route (or follow behind by a generation or two).

InTheKnow said...

I think you are misreading Otellini. AMD will most likely survive because they are doing exactly what he said other compainies would have to do. They are forming a consortium with IBM.

Specifically..."There are not many semiconductor companies that are $5 billion or more in revenue. So economics leads many companies to collaborate.

To give you a perspective on how costly it is to keep pushing the technological envelope, just look at TIs decision to go with foundries for development past 45nm. Unlike AMD, they have at least been able to run in the black over the long haul. Now they are throwing in the towel.

The concern about IBM's schedule may have some legitimacy. From my dealings with folks at IBM they are very much academically minded. So their sense of urgency will be more to publish than to push technology into production. I think they will give AMD access to cutting edge research, but I don't know if AMD will be able to move it into manufacturing quickly enough.

If someone held a gun to my head and demanded I predict the future, I would have to say that you will see some sort consolidation in which AMD will ultimately lose its unique identity. I think the pie is just worth too much for anyone to let Intel have it all to themselves.

Anonymous said...

'From my dealings with folks at IBM they are very much academically minded. So their sense of urgency will be more to publish than to push technology into production"

The other thing which you hinted on is IBM is VERY GOOD at research...development and manufacturing maybe not as good. Look not further than their SiLK research which they touted as a breakthrough in low K dielectrics many years ago (SiLK = spin on low K dielectric). The film tanked when they tried to integrated it into a CPU product - very good properties in research stage (which was all done on 1 or 2 layer structures), not so good when there are 6+ (at the time) layers in an actual product.

For this very reason I would not get too excited about their High K announcement - Intel announced it when they had it in actual product and yield (and will be ramping 45nm with it).

IBM announced they can do it but offered no actual performance #'s nor actual integration in a test vehicle or product (i.e can they get INTEGRATED performance and YIELD) As someone who has worked in this area it is very easy to demonstrate basic performance, but very difficult to integrate and get high yield. I think that is why you see AMD saying it looks like best case is late 45nm node (2009/2010) or 32nm node. If you look at initial Cell yields with the Playstation3 debacle, you'll see IBM's approach to production and yield is different from Intel and AMD.

I too see this as an issue moving forward - IBM doing the research is one thing, getting a manufacturable process is another thing and still takes considerable time, money, resources.