A transcription (about half-done as I type this) of the conference call with AMD executives can be found here.
Two key points:
AMD’s number imply a total CPU market of 35 million CPUs market. Not great, but hardly a disaster: down roughly 5% from 4Q 2000. These numbers imply that Intel sold 2-2.5 million less CPUs than it did Christmas quarter, again, not great, but hardly terrible news for Intel. A big chunk of that is just normal seasonal variation.
AMD got at average of $75 per CPU, a drop of 15% Not great, but the important point is that AMD can basically break even overall at $75 a CPU.
That being said, the big question is: What happened to Intel’s ASPs this quarter? Obviously, they plunged on PIIIs and Celerons. Maybe not so obviously, those plunges were at least somewhat offset by better sales of PIVs: if you sell five 1.7 PIVs at $350, that’s better than one $700 sale.
The problem Intel has in any price war vs. AMD is that they need a much higher ASP to be profitable than AMD. That would be true even if Intel only made CPUs (a back of the envelope figure indicates about a $120 breakeven figure), but on top of that, unlike AMD, Intel’s other businesses (at least up to now) lose a lot of money. Last quarter, the processor business made about $1.7 billion; everything else lost a bit over a billion.
The losses from those other businesses raised the processor ASP breakeven point for Intel last quarter to about $150.
Since almost all of a drop in processor ASP represents a loss in profit, what may seem to be a relatively minor drop in processor ASP can have a huge impact on the bottom line. For instance, last quarter, Intel’s ASP was very roughly about $170. If ASP had been 10% less, Intel’s overall profit wouldn’t have been $650 million, but more like $150-175 million.
Of course, if the non-CPU businesses just lost less money, that would help relieve some of the strain on the CPU business.
Intel’s statements on the matter seem to indicate a drop in profits of very roughly $200-$350 million from last quarter, but that estimate is not too much better than reading tea leaves and could easily be wrong for a number of reasons. If that estimate is right, though, that would be a pretty good performance under the circumstances, though I doubt the morons on Wall Street will agree.
For the near future, Intel’s expenses will go up because the PIV chews up a lot of die space. This shouldn’t be exaggerated, though. If Intel can average $200-250 ASPs on PIVs, they should be no worse off from an ASP perspective than they are now selling PIIIs. Production capacity and marketing capacity then become the critical factors for Intel.
As I said before, I think what we’ll see over the next few months is Intel effectively raising prices with a more expensive PIV product mix, and AMD even more quietly saying, “OK” and hoping to gain marketshare more on availability due to Intel shortages than on price.