When Push Comes To Shove . . .

There have been growing complaints that AMD is stiffing the retail channel, most lately this.

My reaction to the complaints is a simple question: You’re surprised?

I’ve listened to the head honchos, and especially the head honcho at AMD for years, and one message has emerged.

When Hector Ruiz talks about customers, he’s talking about Fortune 500 companies. Or a big scientific research project. Or HP, or Dell. Or anybody else buying big.

He’s not talking about the little people. Or you. Period. End of story.

Not that he minds your business when it’s convenient, or he’s got extra, but when push comes to shove, he’s going to dance with Cinderdella, not you.

Not the little guy. Not you.

In earlier days, AMD got the somewhat-deserved reputation among the Cinderdellas of not necessarily being able to deliver the goods. If AMD is going to succeed in its efforts to become Pepsi to Intel’s Coke, and live happily everafter, they must kill that impression, so the biggies, and everyone else will get what’s left over.

Given where they want to go strategically, I don’t think they really have any other choice in the matter.

What about the little guys? Well, to be crass, they’re screwed. They can moan and groan all they want about it, and build more Intel systems, but if their customers want AMD processors down the road, for goodness or cheapness or whatever, they’re going to have to get over their huff.

The opportunities for payback are pretty limited (i.e., Intel manages to smack AMD back into submission), but even if that happens, do the little guys want to give AMD a kick, only to face an Intel monopoly?

Do I think Intel would act differently put in the same situation? No. The problem is nobody thinks Intel would do otherwise, and they think AMD should.

There’s this myth that AMD is this underdog company that sticks up for the underdogs of the world. The reality is selling to the little guy wasn’t a matter of choice for AMD, it was more like dancing with the stepsisters before Cinderdella showed up. That was pretty much the case during the Jerry Sanders era, and it’s certainly the case in the Ruiz regime.

This myth needs to die. It never was seriously true in the past, and it’s certainly not true now.

Why are these shortages happening? Cinderdella certainly didn’t help matters, but I don’t think they’re the sole “villain” here.

The major complaint is that dual-cores aren’t available, and that’s not surprising. The market is shifting to dual-core faster than AMD can shift to 65nm. Remember that every extra dual core at 90nm chews up the die equivalent of two single cores, so more dual cores mean fewer overall CPUs AMD can put out.

The answer to that is 65nm, but AMD’s delays in shifting to that hurt, and shifting itself will serve to reduce capacity in the short-term.

Keep in mind what AMD needs to do the next few months. It has to convert and/or bring on line 65nm production at Fab 36 now without affecting production. Yes, at least in theory, they have Chartered to fill the gaps, but Chartered will have to convert, too.

Yet another strain is Turion sales, many of which are turning dually, too.

Another contributory reason no doubt is AMD’s recently touted, “Don’t build it until ordered” policy. Sorry, but all this does is shift inventory management from AMD to distributors and buyers. Sorry, but nobody can completely predict demand three months in advance. Sorry, but I don’t think AMD is going to give Dell or HP the finger if they need a few extra hundred thousand CPUs anytime soon and tell them, “You should have thought of that last summer.”

Finally, we need to point out that AMD is huffing and puffing to deliver just 23.3% of the world’s CPUs, compared to 17.7% a year ago, and even if all their expansion plans work out perfectly, would only get up to around 30% of the market by 2009.

Conclusion?

The shortages will pass as production of 65nm processors ramp up in 2007, but AMD still has its one big core problem:

They have no money.

Practically every problem and difficulty AMD has ever had can be traced back to that single problem, no money. It means they have to do everything on a shoestring, and they can’t throw money at their problems.

For most of its existence, that has meant AMD has been relegated to an average 15-18% marketshare, sometimes a few percentage points more, sometimes a few less.

If AMD continued on that path, they would have eventually gone extinct, simply because of the rising cost in fabs. So AMD has gone on what is essentially a “Grow or die” path, trying to become at least Pepsi to Intel’s Coke, and trying to do it on the cheap.

The problem with that strategy is the “on the cheap” part. Why is AMD shuffling production around fabs like a Vegas dealer dealing? They have no money. Why hasn’t AMD developed a new architecture to replace Hammer? They have no money. Why has AMD had so many problems with SOI, why were they effectively forced to partner with IBM on it? They have no money.

There’s nothing wrong with AMD that a nestegg of $5-10 billion couldn’t fix. If they ever got that, that would be the most frightening event imaginable for Intel.

What AMD really needs is to be bought out by somebody with deep pockets who could toss that kind of money in their direction, and I’ve had a deep-seated suspicion that might be a potential goal of some green people. Not now or anytime soon, perhaps after a couple more good years and a lawsuit against Intel which constrains it.

Might. Not will, not maybe. Might, in a few years.

Ed