Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Petz who wrote (31635)	4/9/1998 3:59:00 PM
From: AK2004	Respond to of 1577920

John the predictions are indeed conservative but you have to agree that with new chips to be released there are some level of uncertainty out there. Regards -Albert

To: Petz who wrote (31635)	4/9/1998 4:27:00 PM
From: Time Traveler	Respond to of 1577920

Petz, Unpredictable Mask Step in AMD's 0.35-micron CS34EX Process: Masks are made from layouts. When Jerry is talking about unpredictable process related to a mask, that points towards aggressive design rule just like Paul had mentioned earlier. The more aggressive the design rule (smaller spacing). The smaller is the die size, and that is great, but at the expense of unpredictable yield. When Jerry said 0.25um process does not use this step, what does it mean? 0.25um does not use that mask, very slick indeed. K6 volume of 0.375K in Q2/97, 1M in Q3/97, 1.5M in Q4/97, 1.55M in Q1/98, 2.1M in Q2/98, 2.7M in Q3/98, 3.5M in Q4/98: These actual and predicted numbers do NOT indicate a yield problem solved, rather an indication of ramp-up in production and going into 0.25um. If the yield problem is overcome, I would expect an exponential increase! You need to do your math again. This time, please try to be a little more realistic. Your standard of being pessimistic is still drastically optimistic in real life. JYW.

To: Petz who wrote (31635)	4/10/1998 12:20:00 AM
From: Paul Engel	Respond to of 1577920

Petz - Re: "AMD is getting at least 140 good CPU's from each 0.25 wafer -- this figure has been verified by at least 3 analysts" Let's review your calculations based on earlier, RELIABLE YIELD Information that you had: {==========================} To: Paul Engel (32681 ) From: John Petzinger Sep 26 1997 9:22PM EST Reply #32981 of 34149 Paul, Re: "they could lose money on every K6 they sell" 8" wafers 31400 sq mm, 162mm die size = 194 die/wafer 80% useable die vs theoretically possible die = 155 die At 40% yield = 62 die (I have on good authority that current yield is above this) AT 84% back end yield = 52 shippable die Average selling price $160 = $8,320 revenue per wafer Packaging costs = $2062=$1,240 Wafer cost = $2300 Gross profit per wafer = $4,780 At 3000 wafers/week = total of 2,028,000 K6's in quarter (the predicted minimum) Now, to estimate the fixed costs, AMD's revenue per wafer for the 3rd quarter was 26 shippable die $200 ASP = $5,200 or gross profit per wafer of 5200-620-2300= $2,280. Total gross profit for 3rd quarter=13 weeks 3000 wafers $2,280/wafer = $ 88.9M. Since quarter will have had a slight loss, fixed costs must be about $90M, this includes all the expenses that ate up AMD's profit, such as ramp up costs. Now, go back to Q4 with $4,780/wafer * 13 weeks * 3000 wafers/week, this gives gross 4Q profit of $186,420,000. Subtract the $90M fixed costs and net profit is $96,420,000. Also subtract an additional $10M for higher Dresden fab costs and $10M for higher advertising and we're left with a $76M profit or 0.36/share post tax. Petz {============================} Kind of a bad prediction, eh John? Care to qualify your current wild guesses? Paul

To: Petz who wrote (31635)	4/10/1998 9:00:00 PM
From: Brian Hutcheson	Read Replies (1) \| Respond to of 1577920

Hi Petz , re. your detailed estimate of K6 production Prior to noticing your post on the 9th April I had been calculating a simpler method of obtaining a rough idea on possible number of K6s :- For every 1000 wafers on .25 micron at 50% yield AMD produces 2M K6-3D Additionally now that .25 is proceeding on a steep ramp of clock speed and since Intel does not have a comparable chip in the Pentium MMX line (higher clock speed and superior architecture) , AMD cannot be price pressured by Intel while Intel maintains its own profit margins with PII . IMHO this means that AMD has much more freedom in pricing the K6-3D resulting in a higher ASP . Lets say AMD maintains an ASP of $175 then for every 1000 wafers/week AMD would generate $350M in K6-3d revenue . The question is how quickly can they ramp production ? I believe now that the yield problems are solved they will surprise not only the analysts but also the pseudo analysts from Intel that appear on this thread . OF course if they seriously ramp K6-3D up to say 3000 wafers a quarter and 6M K6-3Ds then that ASP would fall but so what ? Brian PS I enjoyed reading your far more detailed post