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 : Stock Swap -- Ignore unavailable to you. Want to Upgrade?

To: Doug Skrypek who wrote (15181)	8/7/1998 5:18:00 PM
From: Kachina	Respond to of 17305

The story on SOI and IBM is not that simple. There are two factors here. 1. Current manufacturing process technologies are bumping up against their limits in terms of how many microns of width can be etched on a wafer. This is related to diffraction and the wavelength of light used. Current estimates on full retro of a FAB for a totally new process (such as the one out of Hitachi that they are being coy about) are estimated at $10-$15 billion per FAB. Umm. Big investment for something not necessarily proven at scale. Real big. 2. That means that anything that can speed up circuits without going out of the current process methodology and equipment is worth a lot of money. So - IBM going SOI and combining that with copper interconnects means that they can cut power consumption. That means that less heat is created in the chip. That means you can run it without regrigeration. BTW - you can cool a current pentium 200 down to liquid nitrogen temperature and run it at 800MHz just fine. (Except for bus race problems.) Please note that they are saying that they are adding 10% to the "overall cost" of manufacture. This is an important distinction. They lose on the oxygen implant step - lose bigtime. But they gain elsewhere and they have combined this work with serious strides in minimizing defect rates. Cut into defect rate and you cut your overall cost. For this extra overall 10% they get a 20-30% performance gain. That is worth it. It means you can take an existing 750MHZ chip design and make run on a commercial room temp box at 900-975MHZ. Also - there are many other circuits that this process can be applied to. RF, etcetera. Anything that benefits by lower power and speed. So I would say this has bought IBM something real. They and those who work with them can beat Intel. This at a time when Intel is already squoze into a corner.