Hedging . . .

There’s an article about AMD’s Fab 36 at the Inquirer. There’s some AMD roadmaps (along with a 939/AM2 comparison) at HKEPC.com.

Combine the two together, and what do you get? What you don’t get is a lot of 65nm processors from AMD in 2006, or much of 2007.

Any kind of 65nm production at all will only occur towards the end of the year, and even that presumes all goes well with the ever-tricky SOI formulation.

Look at the AMD roadmap on page 2 of the HKEPC column, and it’s . . . timid. With only one possible exception, all the processors shown could be done today.

The only exception to this is the FX-64, which presumably is a 3Ghz dual core processor, and even there, AMD probably could stretch 90nm far enough using strained SSOI to cover that if needed.

What is, again, noteworthy is that there is no talk about 65nm anytime soon, outside of the “Brisbane” chip sometime in the first half of 2007.

It should also be noted that Opterons are not included in these roadmaps, and it’s at least fairly likely any 65nm chips would show up there first.

What to think? This looks like a roadmap from a company that is hedging their bets, one for whom 65nm isn’t quite yet a done deal. There can be a number of reasons for this: getting enough 65nm equipment, getting the SOI formulation exactly right, perhaps stalling a bit to accelerate some more goodies into the 65nm design.

Whatever the reasons, it’s a conservative plan, and it is better to set one’s public goals low and overachieve if things go well than to aim high and miss if they don’t.

But this is not the roadmap of a company all set to spring some miracle weapon this spring, or summer, or fall, or, if things don’t go well, even winter.

Ed