Re: I take it that you believe Intel's depreciation costs will rise by $1B?
No, I wrote With the administrative, personnel, energy and material costs of running 6 FAB equivalents dedicated to CPUs, vs. 4 this year, plus the ongoing costs of keeping the equipment in all of them up to date, Intel will have some heavy costs next year.
Increased depreciation was a minor factor, listed after the "plus."
By the way, the numbers you listed showed, on an annual basis, Intel capex dropping from $8 Billion last year to $4 Billion this year when the most recent statements I recall expected a decrease from $7.5 Billion last year to $5.5-$6.5 Billion this year.
The bulk of Intel production, last year, was 90mm2 PIII/celerons, and the bulk of AMD production, last year, was 128mm2 Athlons.
The bulk of Intel production, this year, will be 146mm2 P4s (with some Tualatin at (I don't recall 65mm2?), while AMD's production this year will be split between 129mm2 Athlons, and 80mm2 Athlons.
On a per chip basis, Intel will be running a lot more wafers on a lot more wafer lines in a lot more FABs operated by a lot more people using more power and material and requiring more administration. Compared to last year, on a per chip basis, AMD will be running fewer wafers, etc.
I also wrote that their biggest risk is that limited capacity will no longer "save them from themselves" by putting a floor beneath how much market share they can gain by lowering prices. By the middle of next year, they can probably get back to 95% market share, if they're willing to accept ASPs of $25 and losses of $3 Billion per quarter (now that's a lot worse than even I expect, but it's an example of the absurd extreme they'll be able to take a price war to if they want to).
And it may not be that absurd - look at what the DRAM producers did to themselves last year. |
