Tuesday, October 17, 2006

What would have been a good Intel Q3?

What would have been a good quarter for Intel?

It is not a matter of revenues, profits, nor EPS. It is a matter of the effectiveness of the strategies the company is carrying out, 'cos things like the price war have a short term effect on the prime financial metrics of a company but have very long term, "strategic", let's say, effects.

The main strategy that Intel set up for this year was the price war.

Why? Even though Intel damages itself in absolute money quantities three times more than the pain it induces in AMD, Intel has insisted the whole year in trying to undermine AMD's momentum instead of going for the profits, there is ample proof of the price war, motivated to try to revert the market from what it is now, a duopoly, to what it once was, a monopoly.

Thus, the first item in the checklist would be: "Is Intel gaining back market share?"

A second notorious issue is the himalayan mountain of way over four billion dollars of inventories that is "overhanging" from Q3, that has also been steadily growing.

A third issue is how well they are doing with the products that will become the bread and butter of the company in the short term future.

1) Intel, obviosly, must be able to determine whether they are snatching customers from AMD or not, for me it is unconceivable that a company that commands 3/4 of the whole market isn't able to distinguish between a good overall market or gaining back market share. In the conference call, management was asked about that, and they chose to use the rather vicarious excuse to duck the question that they didn't know. For me, that is enough demonstration that they are not really sure that they gained back any market share, they would be bragging. Henri Richard also said recently that it seemed that Intel had relented on the price war. It seems that they are dessisting. Following H.R. comments, it seems that the price war was never effective to curtail AMD's momentum, because H.R. speaks with a sense of relief about it. Of course, the duopoly is a reality in consolidation.

In the report itself, Intel mentions record unit shipments, but nevertheless, lower ASPs. Despite seasonality's expected sequential increases in revenues and profits, a quick comparison of the information reported today:

($ in millions)

Three Months Ended Nine Months Ended
OPERATING SEGMENT INFORMATION: Q3 2006 Q3 2005 Q3 2006 Q3 2005
Digital Enterprise Group
Microprocessor revenue 3,521 4,936 10,751 14,483
Chipset, motherboard and
other revenue 1,425 1,434 3,963 4,249
Net revenue 4,946 6,370 14,714 18,732
Operating income 858 2,164 3,165 6,574

Mobility Group
Microprocessor revenue 2,239 2,331 6,544 6,304
Chipset and other revenue 809 639 2,172 1,722
Net revenue 3,048 2,970 8,716 8,026
Operating income 1,260 1,431 3,362 3,784

and that of Intel's Q2 Report:

Three Months Ended Six Months Ended
Q2 Q2 Q2 Q2
- ----------------------------------------------------------------------

Digital Enterprise Group
Microprocessor revenue 3,338 4,603 7,230 9,547
Chipset, motherboard and
other revenue 1,283 1,398 2,538 2,815
Net revenue 4,621 6,001 9,768 12,362
Operating income 931 2,016 2,311 4,405

- ----------------------------------------------------------------------
Mobility Group
Microprocessor revenue 1,958 2,056 4,305 3,973
Chipset and other revenue 731 566 1,363 1,083
Net revenue 2,689 2,622 5,668 5,056
Operating income 946 1,220 2,101 2,351

tells us that Desktop/Server revenues went up below seasonality and net income went down in absolute terms, despite seasonality, being only laptops really growing sequentially. Despite all the bragging of record units, this tells us something very clearly: In Desktop/Server, where AMD's presence is felt strongly, especially servers, Intel is actually doing very bad, having as last recourse the mobile segment of the market. But what is going to happen once Daamit churns out (Intel)killer chipsets with immensely superior integrated, Vista-ready, "bulldozers"?, that's a sidetrack. What is important is that it would seem that Intel reached record units and lower ASPs only because AMD switched the low-end volume markets for the middle end dual core desktop and highest end multiprocessor servers, virtually ejecting Intel from those segments.

There is also murky financial information, I suggest to go to "Syndrome's" for speculation.

2) Intel wrote off a hundred million dollars of inventory. And guess what happened?: Inventories actually grew!!. The thoughest question of all is what is that inventory composed of: Worthless Pentium4, nearly worthless PentiumDs, Yonahs or CMW?

3) According to Intel, they are ramping fast both Woodcrest and Conroe. Their presence is felt, no doubts about it. But how are they doing?

For me, it is clear that Conroe only managed to displace Athlon64 FX and higher end X2s from the top to the mainstream segment. Not good for AMD to be forced to sell former premium chips as "mainstream". Not good at all to have only multiprocessors as an answer, with not even hints of coming to market with something that may regain the absolute single processor performance crown. But this segment is not so important. What is important is that it seems that the whole AMD world migrated from Semprons to dual cores. That speaks of an increase of AMD's ASPs. We will see tomorrow. Bottom line is that Conroe is a victory, a promising line of endeavour for Intel once they manage to produce them in enough quantities, that, is, as soon as "Copy Exactly!" does not tie their production resources to obsolete products.

Woodcrest and their upcoming quadcores are a different thing, though. Woodcrest and Opterons compete just like Conroes with Athlon64s. But, at least for me, surprisingly the whole world received very warmly the Rev. F. Opterons. Why?: DDR2 memory? nah!, Virtualization? it helped, especially because AMD's virtualization is leaps and bounds superior to Intel's, but no. Ah!, the promise of the true quadcores. You see?, Intel is not able to scale up from the machines with two cores. Not with two processors, not with a single processor that has two dual core dice. Above the two cores, in Intel systems the extra cores only add marginal performance. That is, the third and fourth cores may add the equivalent of one core (50% more performance), and above four cores, the benefits become less and less, exponentially. Because there is a problem of bus contention that worsens exponentially. AMD, on the other hand, may promise super linear performance increases (that is, an AMD quadcore may have better performance than double a dual core). This is irrefutable. Thus, the enthusiastic acceptance of Rev F that promises replaceability by true quad cores make them compelling enough to buy them today and wait a year.

Intel is trying to enthuse the market with unsexy products. AMD is preparing a revolution.

Intel is a big company, and I don't understand the whole of it, but in the microprocessors segments, it had a disastrous quarter.


Anonymous said...

Intel's situation is worsening. If you pay attention, actyally 50% of their production is still 90nm. Out of their 65nm, only a small fraction is Conroe. They sold Conroe and Woodcrest in 3Q06, yet, their server+desktop revenue dropped from $4.9 billion to $3.5 billion. Clearly, Conroe is not saving Intel. Their mobile business was slightly down. The only thing showed some growth is their chipset+MB business.

Anonymous said...

Chicago, thanks for your evaluation!

I heard the CC and was not impressed with Intel. It seems their Conroe/Woodcrests are doing well but are not being made yet as a majority of capacity. They seem almost entirely 65nm on CPU's but many of those seem to be that nasty growing inventory of antiquated chips with the old architecture.

To me the main difference is inventory and growth in servers and mobile. AMD hopefully will say they are moving everything they make and are capacity strained, but relieving that with Charter and Fab36. Hopefully AMD will say they have continued to gain in servers and desktop and are prepared to blowout integrated mobile chipsets and 65nm multicore processors.

Erick said...

Chicago, thanks for your evaluation!

I heard the CC and was not impressed with Intel. It seems their Conroe/Woodcrests are doing well but are not being made yet as a majority of capacity. They seem almost entirely 65nm on CPU's but many of those seem to be that nasty growing inventory of antiquated chips with the old architecture.

Eddie said...

Erick: AMD is preparing integrated graphic chipsets vastly superior to anything Intel produces, only second, if at all, to nVidia's.

Being as it is now that the weak component in laptops is not the processor but the integrated graphics, and that it is extremely difficult to change the GPU in a laptop, AMD may topple the whole situation in an eye blink, perhaps as soon as Q1 '07.

I was more than skeptical about the ATI purchase, now that it is a fact I have been able to see that it has the potential to become a fatal wound in Intel's business denying them the relief of mobile