A Year of Fratricide . . .

Dailytech shows an Intel roadmap indicating that in Q1 2007, Conroe will have 40% of Intel’s production.

This is a little slower than historic Intel rampups (the troubled Prescott and derivatives notwithstanding), normally a new generation hits about 50% of total production three quarters after introduction, but not terribly so.

The real problem Intel faces the second half of the year and early next year is that Conroe will have two competitors: AMD and . . . Intel, and it might beat up both of them.

Let me put it this way, what would you rather buy, a Conroe dual-core or a Pentium D dual-core? Practically all of you would choose the first, Intel has to make most people choose the second.

Under normal circumstances, the way you get people to do that is to price the new stuff a lot higher than the old, but the lower-end Conroes are pretty reasonably priced, and not much more than the Preslers.

GHz Matters, Except When It Doesn’t

To put it mildly, selling both is going to be a tricky proposition.

On the one hand, many people will think that GHz matters, so, left to their own devices, they’ll see, say a 3.2GHz Presler and a 2.13GHz Conroe and say, “the Presler is better because 3.2 is more than 2.13.” Intel can certainly keep quiet and let people think what they think when it’s to their advantage.

But then how do they sell Conroes? The sales pitch for those has to be, “Well, this 2.13 is better than that 3.2,” but not too loudly or persuasively, otherwise everyone will buy the processors Intel doesn’t have to sell, and shun the processors Intel does have to sell.

It’s a tightrope. It’s not inconceivable Intel can walk it through the second half of the year by doing things like selling Conroes in only loaded, higher-end systems, leaving the Pentium Ds for the Everyman systems, but if Intel is having to hand out incentives to get resellers to flog Pentium Ds now, how much more will that be the case when they have intramural competition?

Even if things go well, Intel will probably have to sell most of these duals to the OEMs for rather less than $200, and what will that do to Intel’s bottom line, much less AMD’s?

Recently, I went to one of those computer shows, first time in quite some time, and much to my surprise, I found as many or more vendors selling sub-$200 Pentium D 2.8s as AMD chips. Somebody must be buying them.

AMD is in a curiously restricted position competing against the Pentium Ds. On the one hand, they have a better product. On the other hand, they really don’t want to make a lot of X2s until they get to 65nm, simply for capacity reasons. They’ll take the relatively small percentage of people who’ll pay $100 more for an X2 system, and hope most will stick with single core.

If you want to know a sure sign AMD has really increased production capacity, watch the price of the X2s. If AMD starts matching Pentium D pricing, it’s a sure sign they have wafers to burn.

On the other hand, if the consumer mindset shifts to dual-core towards the end of the year, AMD may have a lot of single cores no one wants, and that’s more likely to happen if Intel slips off the tightrope and finds that the only way they can sell their Ds is to sell them at a firesale price.

So it’s a very odd predicament both Intel and AMD are in. It is in the best interest for both of them to stay quiet and not make any sudden moves, because if either does, a price war is the most likely outcome.

How likely is that to happen? Well, on the Intel side, if enough people realize that Conroes are betters than Ds, you’ll have shortages of the first, and fire sales of the second. If AMD does bring a lot of extra capacity on line and starts using it the next few quarters, they’re going to have to sell it one way or the other.

Personally, I don’t think AMD will. I think Fab 36 will make relatively few 90nm chips (they’ll focus on getting 65nm ready), and whatever they make will be somewhat offset by reductions to overloaded Fab 30 production. Chartered probably won’t make a ton of chips in 2006, either, maybe an appreciable amount for the Christmas quarter.

I think AMD overall will end up making somewhat more chips than they are today (maybe a couple million more a quarter), but mainly to relieve shortages, not much more than that.

The second half of the year is going to be a pretty unstable situation. You have two gunfighters poised against each other, and both have very good personal reasons not to shoot.

Will either twitch, for whatever reason? That’s the story for the second half of the year.


Be the first to comment

Leave a Reply