Advanced Micro Devices Inc (AMD)

[alan81]

let's try some numbers...
I think we have more information now with a breakout of GM and units for 2005. If we assume Q4 continues on the current growth pace and hits $1.2B total 2005 revenue comes in at $3.6B. With a gross margin of 55% that puts cogs at $1.6B and gross margin dollars at $2B. From the slide we see units of 42M whch gives us a cost/unit of $38 and an ASP of $86. That all seems pretty reasonable.

Now we have to make some assumptions about 2006. From the slide, they can build 55Mu in fab 30+36. If we assume some ASP upside to $90 we get revenue of $5B. Now the tricky part is figuring what the product cost will be, especially if fab 30 depreciation is dropping. Note that product cost using a NEW factory is about a third depreciation, while obviously for a fully depreciated factory it is zero. We know AMD has recently installed some new stuff in fab 30 for 90nm so 2006 depreciation will not go to zero... nor is it likely '05 depreciation was a full third of the $1.6B COGs. As a swag, let's say fab 30 COGs drops to $1.4B with a reduction in depreciation, or perhaps it stays the same at $1.6B. The even tougher question is what will fab 36 COGs be... If we bound that by say $1B on the bottom, and as equal to fab 30 on the top at $1.6B we get a 2006 COGs range of $2.4B to $3.2B... a pretty wide range actually.

If we track this back to product cost we get a product cost range between $44 and $58. I believe the 2005 product cost number of $38 is almost all single core 512K cache chips. It seems reasonable that if performance needs to shift upward to dual core or a richer mix of 1M cache product that cost would increase this much.

Bottom line of this is total margin dollars could increase from $2B to $2.6B, or it could drop to $1.8B... depending on which cost assumptions you go with. Either way, I don't see it as a huge change from the 2005 numbers, and gross margins will indeed drop to within the range of 40% to 52% depending on which assumptions you make. Of course you are free to make much more aggressive assumptions if you want, but I think these are pretty reasonable.
--Alan