6.15.2007

Intel Breaks AMD's Business Model

It is becoming more apparent that AMD's days as an IDM are numbered.
Forbes:
"Covello told clients he upgraded his rating (for Intel) because AMD's likely move to an outsourced business model will create significant benefits for Intel over the longer term."

An increasingly larger chunk of AMD's volume is no longer generating enough margins to warrant in-house production. Thanks to AMD's price war and failed campaign to grab 30% market-share, the commoditisation of x86 processors was accelerated. AMD miscalculated the large performance gap Intel was able to establish with its Core2 generation of processors. Along with its cost advantage, Intel managed to protect itself from AMD's scorched earth strategy of flooding the market with sub-$100 processors.

Intel understands the danger of having a significantly sized competitor which is why it is determined to continue to put the hurt on AMD. Intel's planned price cuts next month shows its willingness to sacrifice margins in the short term to deny AMD the financial means to expand its in-house manufacturing capacity. In fact the strategy worked so well that Intel even managed to deny AMD the financial mans to run 2 Fabs on a day-to-day basis.

AMD's earning projection shows red until the end of 2008. In order for AMD to survive, it needs to outsource. You can mark this down as the beginning of the end for AMD. No semiconductor company has yet to return from outsourcing. Commoditisation doesn't allow it. Talk about shooting yourself in the foot.

9 comments:

Tom said...

All hail Hector!

I wonder if Barcelona is REALLY delayed and the stepping is so broken
AMD may be shifting to a survival strategy.

I wonder why the $2 billion loan? They must be forecasting losses much farther into the future than anticipated.

Anonymous said...

Unfortunately outsourcing won't work (or work much less well) for a variety of reasons:

1) The foundries that AMD could outsource to are at best the same of not further behind Intel on technology nodes than AMD. This would put AMD at the whim of the foundry's technology node timing and left hoping that the foundry can keep up with Intel - what if the foundry just said from their own business perspective that it was no longer cost effective for them to do 2 year node scaling and slowed this to 3 years (like some folks are starting to do).

2) AMD will have to sacrifice an additional cost advantage to Intel as the cost of an AMD chip would be chip cost + foundry profit or markup. This would allow Intel to squeeze AMD's margins even more with price cuts.

3) AMD designs would be constrained by whatever the foundry's technology boundaries are. For example if AMD wanted to make a part with 80% metal density on a certain metal layer and the foundry's can only meet process specs with 70%, AMD has no choice but to deal with 70% (whereas if they owned the fab/process development they could push development on an 80% process).

Roborat, Ph. D. said...

one major drawback with outsourcing is the loading inflexibility.

Foundry Loading/Capacity allocation is negotiated 6 to 1 year in advance. It doesn't matter if you utilize that allocation or not, you still need to pay for it. Unlike when you have you own Fab you can shut down or change product mix instantly responding as quickly as the market shifts.

Anonymous said...

"Foundry Loading/Capacity allocation is negotiated 6 to 1 year in advance. It doesn't matter if you utilize that allocation or not, you still need to pay for it."

Great point! Chartered Q1 capacity anyone? It was dropped to minimum level of 1000WPSM (which was the contract limit), even though AMD didn't need the capacity in Q1.

Anonymous said...

And in other news, it looks like AMD is following Intel's model:


"AMD pushes new naming scheme"
http://www.digitimes.com/news/a20070615PD211.html

"The first digit of the proceeding four-digit number will indicate the product family; 1000 represents Sempron/Athlon single-core, 2000 represents Athlon dual-core, 6000 represents Phenom dual-core and 7000 represents Phenom quad-core. The second digit indicates relative core frequencies within the product family while the two remaining digits will be used for the future upgrade functions"

Sound a bit familiar to another company? I can't put my finger on it...6300, 6400, 6450, 6600, 6700, 6800...

And why use 7000 for quad and 6000 for dual if they are otherwise the same? Why not just add quad after the name or use the old "X2" and maybe, I don't know, "X4" to represent quad? This way you could completely eliminate a whole series of #'s for dual vs quad desktop, servers,and if ever comes the time, mobile) and maybe simplify things for consumers who may not keep on naming scheme conventions?

"According to sources at motherboard makers, AMD's new naming scheme is more complicated than Intel's Core 2 Duo series and its current scheme. The transition is predicted to take more than a half year to complete."

6 months to migrate a naming scheme?!?! We're not talking migrating products, we're talking about changing the name! seems a wee bit excessive, no?

SPARKS said...

Intel sacrificing margins? Let’s get some perspective here. I understand the point from the good, learned Doctor. He is right. With all due respect, however, Intel is past that stage of the initial AMD, fat margin crush, presently. Before Jan 07 Wall Street thought, as does the good DR., that fat margins (from the old Intel) were the rule. The new lean Intel is now, this date, demonstrating it can and will make money on SLIMMER margins.

Today, from a Goldman Sachs upgrade, INTC rose 1 dollar on the day. Further, it rose over 3 dollars during the entire week. Sachs, and others, are well aware of the price cuts in July. With a strong product lineup and soaring market share, Intel will not only take the cake, it will eat it, too, despite the slimmer margins and price cuts. Think volume and an extraordinarily successful Core2 architecture, whose developmental costs have been more than paid for many times over. Now, till 2008, it is pure gravy out of the bin, be it 4300 or Q XE. Penryn and Nehalem will be the desert, with volume production and FOUR, READ: 4 new 45 Nm Fabs and nuclear control rod metals.

I have positions with Intel (MONEY). I watch these two players, very, very closely. Not only was this predictable, the numbers, with both performance and volume, made it inevitable. Wall Street is taking the new Intel VERY seriously. AMD is a Wall Street laughing stock. Traditional margins no longer apply. Further, if you think Ottelini is going to lay down now, some serious review is in order. Nehalem will be the Coupe De Gras for AMD. He has a plan that is working extremely well. He is going lean with new technologies to keep the pressure on. He smells blood in the water. He is going to ‘Tick Toc’ ALL competition to death with high volume AND slimmer margins. Shareholders (I) will be very happy. This is gospel.

SPARKS

Anonymous said...

"Intel sacrificing margins? Let’s get some perspective here...."

Well stated - the Intel margin sacrifice was in one or 2 quarters - it has stabilized and by measure of virtually any company, other than Intel 2 years ago, it would be considered beyond healthy and good... People keep saying "price war is taking its toll" when really only AMD's margins have been continually falling. Intel's margins corrected once right around the Core 2 launch but they bounced back up the next quarter and have been relatively flat since - somehow with a "PRICE WAR" going on
:)

Out of curiosity, anyone have ideas of how much margin will go up simply due to spinning off the flash biz. I'm too tired to run the #'s. AMD got a very good pop after exiting Spansion - however I think Intel's CPU/flash revenue ratio is higher than where AMD was, so the jump will not likely be as much.

I too have some money vested in Intel and am hoping that maybe the analysts have not factored in the margin pop when flash is spunoff.

I'm hoping that analysts will misunderstand this when forecasting expected margins and when flash is spun off the margin pop may give some upside surprise - though I doubt analysts could be so sloppy, I can dream can't I? Even if it is was only a temporary mistake, it still could be capitalized on...much like when a stock sometimes pops after a good earnings call and the stock drops back the next day as either people are profit taking or people have analyzed things closer and realized they were a little too optimistic with their first reaction to the news.

Anonymous said...

Witness, the fanboi, ripped to shreds!

chang said...

It is nice blog,Thanks for sharing resources related to website design