THC: Get over these numbers with me again, please. You quote a publication stating that a 40,000 wafer starts per month (300 mm) will need 110 DUV steppers. That does not make sense to me. As far as I understand the 200 mm wafers processing rate is 100 wafers per hour (this was a number cited on this thread). Now a 300 mm wafer has more than twice the working area, so let say the 300 mm wafers are processed only at 50 wafers/hour (and thruthfully the number should be better), so even if we assume a very low use rate of only 20 days and only 8 hours per day, a DUV stepper should process 8000 wafers per month. If the DUV is used even 5 times for each wafer that gets us to 1600 fully processed wafer per DUV stepper. 110 such stepper would then process 176,000 wafers per month not 40,000. Either the number of times that DUV steppers are going to be used is closer to 20 times per wafer (very unlikely), or the up time is much lower than 20 days 8 hours/day, or the number of 110 is highly exagerated.
The general rule in business is that when the capital equipment per worketr increases (and the semi must have the highest value probably close to $1.000,000 or higher per worker), the utilization rate has to increase, which means these devices, when up, would work around the clock, so we should use 20 days up, 10 days maintenance and 24 hours per day, thus reducing the required number by another factor of 3). My guestimate is that for a 40,000 300 mm wafer starts, the foundry needs about 10 DUV steppers, 7 of which are up while three are being "maintained".
If over the next three years the 10% of all foundries convert to .25 and lower (120 foundries, which will be closer to 15% of the "real foundries"), the best we can expect should be about 1200 systems.
Any comment is welcome, particulalrly some solid data.
Can we try to clear this together?
By the way, the new figure I am getting for the grand total world wide wafer foundtries in the world is 1200. Of course a bunch of these are antiquated and will be mothballed.
Zeev |
