Advanced Micro Devices
In Reply To:
Barton SOI?

Format for Printing

Format for printing

Request Reprints


By eachus
January 15, 2003

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

It looks like the first version of CH will have 1meg l2, it should have a die size of about 150mm^2, combine that with the cost of SOI wafers and it will be quite a bit more expensive than barton.

The die layout will be very interesting. If AMD has a layout which can be done 3-up instead of 2-up, that will help a lot on costs. But remember that T-bred is 6-up and Barton will probably be 4-up. So twice the litho costs per die of T-bred at 3-up, and twice Barton at 2-up. At 150 sq mm, 3-up that would be roughly 20 mm by 7.5 mm. (And covered by a heat spreader.)

Also at 150 sq. mm. AMD will get fewer candidate dice per wafer than with Thunderbird. (Ouch!) Of course, moving to 90 nm will bring those numbers back into the Thoroughbred area.

Translation: Opteron is going to make a lot of money per chip no matter what size it is. But for Clawhammer to be cost competitive in the desktop arena will take a die shrink. Of course, high-end users will buy Clawhammers anyway, and Barton should carry the mid-range mail for the rest of this year. T-bred will still be around, probably through 4Q03, as a mid-range to low-end part.

Of course, I can't overemphasize that the low-end CPU market ain't what it used to be. For the bulk of desktop users, the difference between a low-end and high-end system is mostly in the graphics and type of CD-ROM, CD-RW, or DVD drive. AMD says its current sweet spot is around 2000+ or 2200+ and I believe them. I expect it to rise to 2400+, but then again, I expected that last quarter. ;-) In any case, even with Barton and Clawhammer, I expect that the sweet spot this year will stay with the T-bred.

What does this mean for important things like profits and stock prices? Opteron is going to be a cash cow. (Intel's Cascade Pentium !!! Xeon didn't earn its Cashcades nickname by accident.) Clawhammer will be a bookmark more than anything, showing that AMD is the desktop technology leader or one of the technology leaders, depending on your point of view. Barton will make money, but I expect that both T-bred volume and profits will continue to be much larger than those of Barton through the rest of this year.

Does this sound like jam tomorrow and jam yesterday but never jam today? Not really, AMD should make money this year. The paper loss in 4Q02 will probably be smaller than expected, and flash will determine whether 2Q03 is profitable or not. Beyond that Opteron should be a big help. By 3Q03 Opteron may contribute a billion dollars or more in gross profits to AMD's bottom line. Let's say AMD sells 700k Opterons at an ASP of $1400 in 3Q02 that is $1.050 billion in gross sales per quarter! Optimistic? Possibly, but even a very pessimistic 400k Opterons at an ASP of $800 could double AMD's current processor sales.

So this year, flash, Barton, Thoroughbred, and Clawhammer all put together might manage break-even. Any obscene profits will be due to Sledgehammer/Opteron. Next year, flash, Athens, San Diego, and Odessa will all carry part of the load. Look at the pricing discussion we just had. In a 4-way system even $2,000 per processor is a small part of the price and the total cost of ownership. This results in low price elasticity for CPUs and high prices.

This brings us back to the issue of fab capacity, UMC and IBM. It is starting to look like this year the need, if anything, will be for farming out T-bred, not Barton. Can UMC do that? I hope so by now. But late this year/early next year the need will be for Sledgehammer capacity. UMC can't provide that, IBM can. With the 90 nm transition, Dresden should be able to handle the demand for San Diego and Odessa, but it might be nice if AMD needs help with Athens production. ;-)

Become a Complete Fool
Join the best community on the web! Becoming a full member of the Fool Community is easy, takes just a minute, and is very inexpensive.