Six Million For 2004 . . .
XbitLabs reports that AMD will make slightly less than six million desktop Hammers in 2004. All of 2004.
Just to put that number in perspective, total AMD CPU production is a bit over thirty million CPUs in a year.
What does this mean? If these numbers are accurate, that means Hammer won’t become AMD’s mainstream processor until 2005. For practical purposes, Hammer will be introduced into the CPU mainstream at 90nm. These 130nm CPUs are just a prequel to the main event.
What will that mean for you? In all likelihood, it means the kind of high prices we projected the other day for most of 2004.
Why Aren’t They Making More?
AMD is going to have a very odd product mix in 2004. They will either charge an arm and a leg for a processor, or practically give them away with the cornflakes.
And they’re giving priority to the ones they’re practically giving away with the cornflakes.
Imagine someone telling you, “Hmmmm, I can make expensive processors that will give me a big profit, and I can make cheap processors I break even or lose money on. I’d much rather make mostly cheap processors.”
Would anyone sane willingly do that if he really had a free choice in the matter?
Don’t tell me anything about how expensive A64s are to make. Yes, the ones with 1Mb cache do chew up a lot of die space, but what about the ones with 512K? That’s no more cache than a Barton, and AMD isn’t complaining about selling those for $90.
AMD is effectively saying to overclockers, “Give us $400 or give us $50/$90.” Would it not make far more sense to give them a $200 option, and get $200 sales rather than $50/90? AMD would certainly make money from that. The overclocker buying that would most probably buy a 90nm processor in a year’s time, and has effectively committed to the Hammer platform by that $200 sale.
Well, yes, AMD is providing a $200 option, an Athlon 3000+. But what do they do? They trash their own option for the performance people by telling people that it’s going to be downgraded and sent to the bargain basement by the last quarter of the year.
What are these people doing?
Not even I can believe the corporate execs at AMD are quite this stupid/incompetent. Circumstances must be forcing them into this approach. We pointed out a number of possibilities why this might be the other day.
Whatever the exact reasons may be, for practical purposes Hammer has been delayed as AMD’s mainstream processor for at least another six months.
$400 . . . . Later
On Monday, we’ll report on this more thoroughly, but we asked you about what you thought about all this high finance, you’ve sent two types of responses:
Here’s the first type:
“I agree that this is AMD’s policy, but disagree that it is a mistake. In fact, I’ve been arguing for several months now that AMD is positioning itself at the high end of the x86 market, and leaving Intel to have the low end. This is a reflection of Hector’s stamp on the company – ship performance with quality and make money. Servers, high end desktops and desktop replacements – that is the new strategy.
“Keep in mind that AMD will be offering a full lineup of K7 product this year, and that will be available for the folks who want a cheap system. You seem to want a top performance system, and you want AMD to lose money delivering it for you. Wake up and smell the coffee! Those days are over. You won’t get it on 90nm, either, although I expect more product in the market’s middle-priced sweet spot then.
“When will AMD provide 64-bit product for the low end? When they get 300mm/65nm in full production in 2006 (and the production costs justify low price). Until then, the new AMD is a supplier of high end CPUs for those of us who are willing to pay a fair price for the best technology available.”
AMD likes people who think this way.
But then we have the second type:
“Personally, I’m going to sit on my new Barton 2500 socket A A7N8X Dlx. For $200 (cpu & board) it booted at XP3200, is running XP3400 (205 FSB) at stock voltage. When XP3600 or better comes out, guess I’ll up the voltage a tad until I run out of juice and patience at 215-220 FSB. Then sit tight and see what 2005 brings.”
Unfortunately for AMD, for every person who says more or less the same thinga as the first person, I’m getting around twenty who are basically saying what the second person says.
I’ll grant you, our audience is probably a bit more inclined to . . . thrift than the norm, but the percentage saying, “Thanks, but no thanks” is so high that it’s hard to imagine that it will be terribly different elsewhere.
So if the OEMs don’t terribly want it, and the bulk of enthusiasts don’t want it, just who is going to buy this for most of the year? How much money can you make from an unsold processor?
Seems to me the people who’ll end up buying this are the Maximum PC type. I don’t mean to knock them, and if AMD can sell them all the Hammers they can make, God bless them, but are there enough of them?
And what happens if there isn’t?