Elmer, re: Fab capacity, actual output, demand, margins...
Without wanting to get too involved in this (at least) 5-year argument (I distinctly recall this discussion from when I first started reading SI in '99), I do want to make a subtle point that I believe is being overlooked:
It is actually easily possible to be both demand-constrained and still be able to sell many more processors (at same prices)!
The reason is simple: AMD could well be demand constrained at one end of the performance spectrum, but nowhere near that at the other. This argument would work for either end, but for the sake of argument, let's say they are demand constrained at the high end.
AMD has to make money on every wafer that goes through the factory. But the resulting processors bin roughly according to a Gaussian distribution. In other words, there will always be far fewer high-end chips than low-end chips.
In order to make money, AMD needs to reach a certain "wafer ASP". Low-end chips would sell for less than the average, high-end chips for more. (No kidding).
However, if AMD is demand-limited at the high-end, they are suddenly faced with a problem of satisfying demand at the rest of the spectrum without losing money; even if they are charging the same prices for these parts as before, they would lose money, since there would not be a similar sale of high-end parts to bring the total "wafers ASP" up over break-even.
As it pertains to the current discussion, this presents an easy, elegant way in which what AMD has said can be reconciled with what we see (and your BOTE calculations).
-fyo |
