AMD announced another slightly profitable quarter yesterday. CPU sales dropped slightly from the previous quarter, and still remain almost 20% off AMD’s peaks from a few years ago.
The specific financial details aren’t of much interest except to verify one key fact.
AMD isn’t selling many Hammers.
The conference call after the earnings announcement is littered with attempt after attempt by financial analysts to try to find out 1) how many Hammers were sold and 2) how many Hammers did AMD plan on selling for the next few quarters.
These are hardly unreasonable questions, but AMD simply refused to answer the questions. They said that Hammers sales doubled over the quarter (but we didn’t and won’t tell you what those were), that Hammer units were something over 5% of total CPU sales, and that Hammer revenues (though not units) were expected to exceed Athlon revenues in Q4.
However, one number AMD can’t refuse to disclose is their overall CPU revenue. The quarter that Hammer really starts selling, CPU revenues are going to jump up a lot, simply because Hammers overall cost a lot more than Athlons.
If you play with the CPU revenues and ASPs estimates for Athlons (which have pretty much held or gotten a bit better the last couple quarters), you come up with a guesstimated figure of about six million Athlons sold for about $80 ASP, and about 300-350,000 total Hammers (Opterons, FXs, Athlon64s) at an ASP of around $300.
These figures are probably off a bit, but I’d be very surprised if they were more than 20-25% off.
350,000 Hammers is hardly a world-conquering figure; roughly 1% of worldwide x86 sales.
Why is the figure so low? A figure this low points to one primary reason: lack of demand.
It’s not lack of production capacity. Three months ago, AMD said it could make over a million of them. This estimate doesn’t even come close to that figure.
A combination of lack of serious OEM interest due to no x86-64 Windows OS and high prices seem to be the main culprits here.
For what reason(s), AMD doesn’t seem to want to make a lot of Hammers until they get to 90nm (which seems to be doing OK, AMD had no problem being specific when talking about that.
This ought to make you think twice about wanting to buy one until they do get to 90nm and get serious about selling them, and it ought to make you laugh when you read accounts about these chips taking over the world.
The numbers aren’t there.
Such talk is just fantasy. Sheer fantasy.