The Real CPU Battle of 2004

Add Your Comments

AMD posted a loss of $140 million last night. This is actually better than expected, since AMD managed to cut costs more than even they expected.

For more details, you can read Chris Tom’s notes on the earnings conference call.

Really, all you can say is that they’re betting the company on Hammer: that they can make and sell them in the millions, and sooner rather than later.

For the next year, the question will be “Can Hammer revenues more than offset the inevitably deteriorating revenues from Athlons and get AMD into the black?”

Look into that question more closely, and it really becomes “Will AMD fans pay a lot more for Hammers than they have for Athlons?”

That is AMD’s core problem. Most people buy AMD processors because they’re cheap. When they stop being cheap, they stop buying them, and “cheap” has become a lot cheaper the last few years.

That’s the trap AMD is in, and that’s why I’ve been gloomy about the company’s prospects. If they don’t get out of it, they’ll be history.

Short Term

AMD has two kinds of financial problems: short-term and long-term.

The short-term problem is not running out of money before Hammer gets established. There’s two aspects to this.

First, the initial generation of Hammers will offer little if any performance improvement over current Athlon systems. When we asked a while back, over 90% of those who responded said they wouldn’t be interested in Hammer until it reached a 4GHz (equivalent) level. That effectively means enthusiast sales will have to await the 90nm generation of these chips.

Second, once the performance level is met (by which time AMD will have burned much if not most of its remaining cash), the established user base is likely to have sticker shock when they see Hammer pricing.

Whether it’s Athlon or Athlon64, AMD cannot survive in the long term getting an average $60 a CPU (which is what they’ve been getting most of the time the last year or so). They really need something approaching $100 to be financially secure. If those numbers don’t go up, AMD is history.

The real battle in 2004 isn’t going to be AMD vs. Intel. It’s going to be AMD vs. its current customers. It will go something like this:

AMDers: “I’m not going to pay a lot for that CPU.”
AMD “You’ll have to pay a lot more for that CPU.”
AMD “We’ll go under if you don’t.”
AMDer “No, you won’t.”

Talk about mutually assured destruction.

You can’t reason with “No, you won’t.” The only answer is “No, you won’t go under” is to go under, and then it’s too late.

Unless AMD can attract an entirely new audience that is willing to pay a good deal more for that CPU, their current customers will destroy them.

That’s the bankruptcy/takeover scenario for AMD: they price high and their customers hold out for low. Then IBM takes over, turns AMD into something like Alpha and there will be no more talk about $50 processors.

And then the AMDers will blame anyone and anything except themselves.

Long-Term

AMD’s long-term problem is generating enough profits to pay for the billions and billions of dollars in fab costs necessary to keep up with Intel.

You can say, “Well, IBM can do that for AMD, and you’d be right.” However, that’s just paying more a different way. IBM will charge AMD plenty to fab their chips for them (how else are they going to make their billions invested back?), and rest assured IBM will make a profit on the deal. That will drive up AMD chip prices, too.

Cause of Death: Don’t Charge You, Don’t Charge Me, Charge The Guy Behind The Tree

When I’ve written on the subject before, I’ve gotten plenty of emails essentially saying, “Not my problem. Let AMD find somebody else to pay a lot.” People grossly exaggerate the impact Opterons can have on the bottom line. Or they think IBM is some fairy godmother.

The problem is there is nobody else that can generate enough revenue for AMD. It has to come from the desktop processors. Opterons can help a bit; they can’t do it by themselves.

These people will defend AMD to AMD’s last dollar. With those are your friends, who needs enemies?

We’re not talking about a lot of money here. We’re talking about something in the neighborhood of fifty dollars between being financially terminal, and decently profitable.

The great irony is that unless there’s a major attitude adjustment, a whole lot of AMD fans will become Intel’s biggest allies, and find themselves holding out for blue.

Think about it.

Ed

Leave a Reply

Your email address will not be published. Required fields are marked *