AMD has all kinds of problems, but it isn’t like Intel is doing so great, either.
Its 45nm quadcore Penryns are, well, scarce. They aren’t available in a lot of places, those places that do have them don’t have a lot of them, and they’re charging a premium for the ones they do have.
Fudzilla, commenting on this, just said, “We believe that Intel has bigger issues with the 45nm quad core manufacturing than it wants to admit and its time to raise this question.”
OK, fair enough. While we won’t preclude some inability to make the chips, there are two other factors which could explain most if not all of the shortages. Unfortunately for Intel, these reasons indicate that problems will probably continue for some time to come.
Intel never planned on making a lot of quads Previous Intel roadmaps showed that quadcore production would constitute a good deal less than 10% of total production by the end of the year, and the percentage for around now would be more like 5%. These estimates were made before the six week Delay That Wasn’t A Delay, so it’s likely that 45nm production is falling short of even that. So it’s not too surprising that demand would exceed supply.
You might say, “Well, should Intel adjust its production accordingly? After all, all they do to make a quad is slap two duallies together, that shouldn’t be too hard to do.”
This is a very reasonable point to make, except that . . . .
Intel Is Short On Fab Intel’s normal procedure for new process generation is to begin making them at their R&D fab in Hillsboro, then start production at a few fabs, usually one at a time, a number of months apart.
This has also been true for 45nm chips. The first were cranked out at the development fab, and the first “real” 45nm fab (Fab 32)was opened in Chandler, Arizona last October.
The original schedule called for a new 45nm fab (Fab 28) to be opened in Kiryat Gat, Israel in H1 2008, followed by a retrofitted fab (Fab 11X) in Rio Rancho, New Mexico in H2 2008.
Well, something happened to that plan. Take a look at this February, 2008 article from Reuters and you’ll see that the Israeli fab now won’t open until H2 2008. That date came from the general manager of Intel Israel, who presumably knows what she’s talking about.
While this delay may not be the reason for the current shortage, it will certainly ensure that such shortages are going to last at least until Fab 28 is up and running.
And it’s going to have to do that pretty quickly at the beginning of H2 if Paul Otellini’s pledge of 45nm crossover in Q3 is going to happen.