Obviously, Intel has had all kind of problems the last three months. Despite that, they still made $1.8 billion. The financial ramifications of Prescott and a perhaps slow rampup of socket T platforms will only play out in future quarters.
Still, when you make 1.8 billion dollars, you have a pretty big cushion. The next six-twelve months may not be too kind to the stock price, but it’s not like Intel is going to get into any real trouble as a result.
The one item of interest in the Intel financial reports is that Intel isn’t saving as much money as expected (at least not yet) with 300mm wafers making 90nm chips, but again, that’s more financially than technologically interesting.
All that talk about “best quarter yet” has nothing to do with CPUs. AMD took 100% control of the flash memory business they previously shared with Fujitsu a couple quarters ago, so their revenues compared to a year ago were bound to go up (plus, they are doing pretty well in that area).
Overall, AMD’s CPU revenues were down just a little (which is to be expected for Q2 of a year), but gross revenues hide the real story.
What has been happening since the beginning of the year is that XP revenues has been dropping quite a bit (mostly due to lower prices). That’s why AMD’s CPU revenues have been flat the past few quarters when they “should” have been jumping upwards. They’re getting more money from Hammers, but less money from XPs.
I’ll go into more detail in a future article, but if you read AMD’s SEC reports together with market research estimates very closely and see what they say and don’t say, this appears to be the case.
In yesterday’s conference call, AMD sort of acknowledged that. They said that Hammer revenues were up 50% and XP revenues were down. After all, when your total CPU revenues are down a little, and your unit sales were up a bit, and you only make two kinds of CPUs, if one type goes up, the other has to go down.
What does that mean? On the one hand, AMD is probably selling more Hammers than people like myself have been estimating.
When I had estimated 350K Hammer sales the first quarter, I had presumed XP unit revenues had remained steady. That appears to have been an incorrect assumption. It looks like Hammer sales were up a lot the first quarter, pretty much offset by a sizable decrease in XP unit prices. The same appears to be true for this quarter. Total Hammer sales this quarter probably were close to a million units.
On the other hand, AMD is getting rather less for an XP than they did six months ago. Back then, they were probably getting about $80, now, it’s probably closer to $60.
It probably won’t be too far off to say that AMD made about 6.5 million XPs for about $60 a pop, and about a million Hammers for about $200 a pop.
Of course, any CPU company is going to have the same kind of trend transitioning from one generation to the next, but so long as the rampup of the new generation is fairly quick, this is no big deal. That’s not going to be the case with AMD, which is particularly inexplicable given the relatively large dollar difference between two generations of processors.
Per the conference call, AMD doesn’t apparently expect Hammers to outsell XPs in unit sales for another year. They expect Hammer revenues to exceed XP revenues by the end of the year, but as you can see from their relative ASPs, AMD will probably not make more than 1.5-1.8 million Hammers by the last quarter.
For whatever reason, AMD either will not or (more likely) cannot make a lot of Hammers any time soon. (If I had to bet on the “why,” I would say that they don’t want to have a lot of 130nm Hammer capacity, and they’re financially constricted from ramping up 90nm Hammer capacity quickly.)
So AMD for about the next year will essentially be putting most of their effort into making and selling a severely aging old budget line, rather than having new-generation capacity ready for those not happy with a rather troubled Intel line of products. It would be like Intel getting 80% of their unit sales from Celerons.
This leaves them financially vulnerable. Not vulnerable as in the company going under, but vulnerable enough so that a soft market and/or some Intel pricing games could knock them back into the loss column once or twice in the next twelve months.