Crossroads for Intel

pillageplunder writes "Businessweek offers a pretty balanced read on what challenges Intel faces in the upcoming year. Rivals Samsung and AMD are making inroads on Intels core businesses, an expected cyclical industry downturn looms next year, and with several critical delays in new (for Intel) markets puts its strategy at risk. A neat read."

Comments

Cyclical downturn?

[tdvaughan]

Could someone explain why the semiconductor industry is 'cyclical'? What is it which makes a downturn predictable, or is it a self-fulfilling thing (lack of investment during predicted downturns causes otherwise unnecessary lack of performance)?

Re:Cyclical downturn?

[erick99]

I've been on the marketing & sales side of computers for 21 years (started in Oct '83) and the only cycle I've ever been able to reliably predict is a slow down in corporate purchasing during the month of February. That does not necessarily relate at all to the semiconductor cycle, if there is one. I used to meet regularly with Intel marketing reps and they never mentioned a cycle. There may well be a longer term cycle, such as a four year or five year cycle where so many machines are bought at the beginning of a major product cycle such as the intro of the P4, for example. In that case, a lot of machines would come out of service starting at 2 years (leases) and out to five years (fully depreciated). This is all to be taken with a grain of sale of course - there are just too many variables (intervening and contravening).

Re:Cyclical downturn?

[Anonymous Coward]

Most businesses are considered "cyclical". Basically, this means they do better whent he economy is better, worse when the economy is worse. i.e. they do better when people and businesses have more money to spend. People hold off buying when they don't have the money, because these are discretionary expenses.

This is in contrast to some businesses which are fairly "noncyclical"--demand is relatively constant over time, regardless of ability to paty. Medical care is a classic example here--people don't decide to hold off on having tha heart attack until they have a better job.

Then, there are some "countercyclical" industries--ones that do BETTER when people have less money to spend. Examples here are businesses that make "inferior" goods--cheaper substitues for more expensive products. They do better when people have less money, because in good economic times, their products are more attractive.

To an extent, there's an aspect of being "cylical" to most businesses, but some are more tied to economic cycles than others. Intel makes a good case for being a very cyclical business--computing upgrades are a fairly discretionary expense, and delaying upgrades is a fairly common response to bad business climates. On the other hand, when the economy picks up, and people have money to spend, getting those computer upgrades they've been meaning to get for awhile becomes more attractive.

Re:Cyclical downturn?

[nelsonal]

Normally a cyclical industry is one that involves large capacity additions which make up both a signficant amount of the productive capacity and a signficant portion of the cost of each unit. Semiconductors and tankers are both classically cyclical industries. In real terms let us imagine that you and I have built a brand new fab, we perhaps raised $500 million in equity, got grants of $500 million and borrowed $2 billion. Perhaps we decide to compete in memory manufacture. Prior to our fab going on line let us say that capacity was 100/k 8" wafers per month. Following our fab now capacity is 140k 8" wafers per month. Each month to pay for our fab we need to generate sales of about $1600 per wafer to cover our fabs cost. If we did our market research well we should be able to cover that cost if we didn't (perhaps price declines signficantly at 145k wafers and one of our competitors increased their capacity by 20k wafers while we built our fab. Now everyone is producing to shut someone down (or waiting for demand to grow so that we can produce 160k wafers profitably (which could take several years). Upturns are caused by longer waiting by all competitors downturns are caused by capacity being added more quickly than demand, and the most money to be made is accomplished by being the only one to build before an upturn. Under those three rules (with no collusion) you will always have big upturns then overbuilding then downturns then upturns and you have a nice sine wave cycle. The overall trend of the sine wave is up, but the sinewave has a big amplitude.
In processors and other high value chips this is less true (as engineering and design talent reduces the number of competitors you are dealing with allowing more normal capacity increases to take place. Over the last thirty years only three big semiconductor companies have reliably generated more cash than is required to build for the next cycle (Intel, Analog Devices, and Linear Technologies) all three compete with very few competitors and their products require a considerable amount of design before manufacture can take place.

Re:Cyclical downturn?

[supergiovane]

You're Mr. Intel. You have to explaine to your shareholders that AMD got it right and they are now driving innovation (e.g. x86-64) and now you're the one who has to catch up. What would you say?

1. We're in deep shit, boys! You'd better invest your bucks in another company.
2. It's the cyclical behaviour of the semiconductor history. Now we're getting hit, but next year we'll kick their asses and we'll reduce them into dust. So, don't worry and give us your money.

Maybe I'll do my part next year...

[jmcmunn]

And buy a new processor to upgrade my 300Mhz PII I am running here at home. Nahh....it still loads Slashdot just fine. I'll wait till the next generation come out and then buy one of the current chips. (I have been saying that for 4 years now)

Re:Maybe I'll do my part next year...

[joib]

I would like them to only release chips within a regularly defined cycle of say 500Mhz speed increases rather than release every improvement they can squeeze out of the chip. I think people would find it easier to plan and commit to a purchase this way. I think processors are fast enough now to handle the needs of the vast majority and theres not a great deal to be gained by flooding the market with differerent processor speeds and people _always_ waiting to maybe purchase the next small incremental release.


Umm, no. That wouldn't be a very good idea. The reason, in short, is price discrimination. By having a wide variety of products, they can better milk the customers. And the customers win too, since they can choose which product best matches their requirement. It's a win-win situation, so to speak.

Is it a neat read...

[Ghoser777]

because it paints a major decline in the Intel empire, or because it actually has insightful commentary and information?

And yes, I didn't RTFA.

Matt Fahrenbacher

Re:Is it a neat read...

[steevo.com]

It is a neat read because I don't work at Intel anymore.

Competition is good

[ancice]

Competition is good. At the worse, if it doesn't accelerate the progress of better products, it will at least create a check on the dominant players.

Although Intel is lacking on the 64bit side.

Hmmmm

[antifoidulus]

problems at Intel, problems at Microsoft. Could it be that some companies just get too big for their own good?
Happened to other comanies, just look at US Steel, in 1918 they represented 3% of the GDP of the US, but they got too big, and eventually competetitors, both at home and abroad ate up most of that.

Re:Hmmmm

[Anonymous Coward]

Not really comperable. Steel is a generic product where there's little brand loyalty, and essentially no product differentiation. Microsoft and Intel, on the other hand, make products that no one can make directly--you can compete with similar products, but some shop in Taiwan can't turn out Pentium IV's.

Re:Hmmmm

[cyngus]

I think that's pretty much the story of corporations through all time. I think that extends even beyond corporations, to countries.

Let's look at Intel and Microsoft. Both rose to dominance because they had a good product at the right time, with good marketing. (I'm a Mac fan, I think windows is sh*t, but there's no denying that Microsoft has made computers more accessible to a wider audience, although Apple has always made the better product.) Now both are having some problems, why? Three main reasons:
1) Everybody targets the leader - if you're the leader in an industry everyone can see your weaknesses and target them to take you down. You're the guy to beat and people are going to try to do that.
2)The leader is big, and knows it - the leader of an industry is typically big, has big sales, big profits. They spend accordingly and build out accordingly, adjusting to lower profits is harder when you're used to them.
3)The leader is typically slower - 3 follows from 2, in that if you're a bigger company its harder for you to change course and take advantage of new ideas and trends. Firstly, your organization is larger and therefore harder to manage. Secondly, your customers tend to hold you more accountable to servicing them, the underdog gets more leeway, because he's the underdog.

So companies tend to start out small, grow, become too big to adjust quickly to a changing environment and then die or breakup. Some companies (IBM is a good example)manage to just fall into decline for a while and then emerge as a power player again, but this is hard to do for several reasons such as regaining customer confidence, having enough money to engineer the turn around, and the difficulty of changing the corporate culture to fit the reinvented company.

Still...

[StevenHenderson]

As long as Dell is almost exclusively Intel, then they ought to be just fine. It is Intel's exclusivity agreements that will sustain them in times like these, I'd wager. (Yes, I know Intel's problems aren't just in the desktop market, but I like to over-generalize).

Xscale

[mirko]

The i86 architecture is dying and Intel could not release decent 64 bit proc at time, so, they'll have to rely on the Xscale processors which are, after all, ARM compatible.
As the ARM has had the hugest sales in the world during the last years (not on the desktopm, but everywhere else), this'd just imply that Intel will keep its domination but outside the PC market.

Intel aren't doing that badly in other areas

[lukestuts]

I hear the pipeline on the P5 is going to be so long that Halliburton want to license it for reconstruction work in Iraq.

Lack of vision

[Anonymous Coward]

Intel has made some pretty big mistakes over the recent years, in some cases going against common sense:

RDRAM, Itanium, 64-bit extensions for x86, frequency as sole measure of performance, ...

It should be no surprise that now Intel's future is clouded. They have no one to blame but themselves.

Re:Lack of vision

[MtViewGuy]

The biggest fiasco for Intel was the Itanium project, which showed while it was a technically-excellent CPU it also exposed the big problem of lack of software to support the CPU.

Meanwhile, AMD brought new life to the X86 architecture with a modern developed from scratch CPU design using the Athlon CPU core. Note that AMD's CPU's have truly impressive performance per CPU clock cycle, and AMD's decision to move the memory controller onto the CPU die with the Opteron/Athlon 64 CPU's allows AMD to match the performance of the latest Intel Pentium 4 CPU's without Intel's need to run very high CPU clock speeds.

One word.. Inventory

[tanveer1979]

The Semiconductor industry inventory stats are ones which give CEOs nightmares. Actually semiconductor companies mostly make chips which are sold to big buyers like Siemens(modems), ENI(modems), DELL, Nokia etc., So these companies tell the semiconductor guys I need so many chips for so many cell phones/cameras etc., And these guys over order, by a small margin. Next year new tech old models are sold for scrap and again inventories rise. In many cases over ordering reaches levels which are uncomfortable. You cant throw away 20% of your cell phones at cost price, can you? So they dont order. But Semiconductor companies have huge Fabs running. And when such cases arise you have fabs operating at half capacity or even lower. And this leads to big losses. Another problem is that a new chip comes from design to fab about 6 months before production begins, and if problems come in the chip it may actually see the vendor after one year! So what do we have here. Based on demand this year, we plan for next year and if inventories pile up its bad luck.

If you wonder why cant semiconductor companies reduce production, the reason is that when we come out with a chip, ie design a chip there a minimum number which is required for the chip to be profitable. This number is in the range of 500000+ units. Such things are hard to predict. In case of a DSL/cable modem chips the design and conception start one and a half years before release to fab. And six months after that full blown production starts. So we have to know 2 years in advance what to do people want. Its 2 years of R&D by over 100 engineers which leads to a chip. And look at the infrastucture investment. Farms of 100s of 64bit 2GHz+ machines, Ultrasparcs etc., running for 1.5 years simulating, testing, designing.

Get the idea? Chip design is a costly business, and unless market analysts get more accurate instead of being stupidly bullish, this cyclic downturn may be much softer

AMD dualcore Opteron

[geeveees]

In other news, there are some benchmarks on AMD's dualcore Opteron: http://episteme.arstechnica.com/eve/ubb.x?a=dl&s=5 0009562&f=174096756&x_id=1097194717&x_subject=Opte ron+dual-core+details+emerge&x_link=http://arstech nica.com&x_ddp=Y/

It appears AMD designed the Opteron from ground up to be dualcore.

x86 architecture still alive thanks to AMD.

[MtViewGuy]

Actually, the x86 CPU architecture is still alive thanks to a company called AMD. :-)

AMD's groundbreaking Athlon CPU core is far superior to what Intel has, and the Athlon XP showed that you don't need ridiculous clock speeds to get superior overall CPU performance, thanks to the the combination of the very efficient Athlon CPU core and generous on-die L1/L2 CPU memory caches. AMD's decision to put the memory controller onto the CPU die with the Opteron/Athlon 64 CPU's also demonstrates how to get superior CPU performance without running high CPU clock speeds like Intel needs to do with the Pentium 4 CPU's.