To: Elmer who wrote (4120 ) 1/31/1998 7:36:00 PM From: Paul Engel Read Replies (4) | Respond to of 6843
Elmer - Re: "The tens of thousands from SDC is my point. MAYBE, once again AMD has a process working at SDC and can't transfer it to Fab 25." AMD said they shipped 10,000 0.25 micron K6 CPUs from the SDC - NOT TENS OF THOUSANDS! Let's do some back of the envelope calculations. Let's assume that AMD was cranking these out over the past 8 weeks of the quarter - with lots put into the line 13 weeks ago - at the end of the quarter. Now, let us assume that only 500 wafer starts were allocated to 0.25 micron K6's - the rest (500 - 1500) being used for FLASH development, short loops, or whatever. 500 wafer outs over 8 weeks produces 4000 wafers. IF 10,000 good die were ultimately shipped, AMD must be running 3 1/8 net good die per wafer in their DEVELOPMENT FAB. Now - do you see why AMD may have just a bit of difficulty transfering this process to Fab 25? By the way - Ashok Kumar - an analyst who predicted the AMD/Compaq deal, Intel's blow-out Q4 earning - also stipulated that AMD was yielding 1% on the 0.25 micron process. 1 percent = 1/100. At about 400 physical die/wafer, this 3 1/8 die per wafer that I calculated is VERY CLOSE to 1%, or 4 net good die per wafer. I think AMD has a real problem on its hands! Now - at a wafer cost of, say, $2500 - a good approximation - AMD is producing 0.25 micron K6's at a cost of about $625 per unit! I used the higher 4 die per wafer (1% number). Think of that! AMD is selling these for about $281 - which is an UPPER estimate based on 25% below Intel's $375 for a 266 MHz Pentium II! That means, AMD is losing approximately $344 per 0.25 micron K6! Can you guess what that will do to AMD's rather RED bottom line? Paul