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.

Technology Stocks : Cymer (CYMI) -- Ignore unavailable to you. Want to Upgrade?

To: Ian@SI who wrote (21034)	2/5/1999 1:02:00 PM
From: Andrew Vance	Read Replies (1) \| Respond to of 25960

I Happened to be scanning some of my favorite threads and came across your question. It seems that Ian has a good handle on the answer but let's try to narrow it down a bit. 1. We have to assume that the 300mm fab is going to have the same wafer output as any other INTC 200mm fab. The only reason for going 300mm is to increase the number of devices per wafer processed and therefore the number of devices per factory. The standard financial metrice used is wafer cost. The less wafers you process, the higher the wafer cost. While yields are important, they are rolled into other metrices. So we have to assume that the # of wafers processed there would be the same, yielding 2.25 more area and probably 2.5 times more devices (larger yielding sweet spot). 2. INTC is a high volume low mix producer so a good estimate might be 4000 wafer per week started which I will guard band down to 3000 starts per week. The manufacturing cycle time also needs to be facotored in (a guarded secret), which determines the number of wafers in the fab at any given time. Considering the complexity of the microprocessors and levels of processing, we will assume we are dealing with a 21 level process, though ti could be more. 3. If we factor in a 2 percent rework rate and a slight line yield issue, your processing guardband rises to 3100. Factor in a 60% critical to non critical level scenario and about 13 levels will be processed on DUV steppers. 3100 * 13 = 40,000 layers per week and divide by 80 WPH stepper throughput and 20 hours per day (utilization and other downtimes like PMs, etc.) 40K/80 = 500 500/20 = 25 steppers While the above is an extremely abbreviated method of determining the number of steppers, it gives you a slight insight to parts of a process used to determine the answer. As we become more aware of the actual numbers, we can come up with more precise answers, and apply the more detailed equations used to pinpoint the number of system required. I have worked on capacity models in a prior incarnation in the industry and would be more than happy to dust off my proprietary program and work the numbers IF and WHEN we can fill in the details of this process and the new fab. The final guardbanded number comes very close to the numbers thrown out by Ian, and should validate his comments. With 25 steppers you would probably have 2-3 spares of lasers on hand, depending on the availability and delivery of replacement lasers. Spare lasers are critical since CYMI may only have the ability to produce, on the average, 2.5 units per day.(1000 units per year was the last capacity number I came across). Some where down the road you should expect at least 12 DUV system per IC manufacturing plant world wide, as everyone will eventually move to DUV. I would imagine that at 0.10u technology, both the critical and non critical levels will be run on the DUV systems. So as time goes on, every DUV user will be backfilling their fabs with more DUV systems as their devices shrink to the next device technology. Andrew Vance - Editor RadarView The Semiconductor Financial Newsletter stepman@hotmail.com