To address your points:
MU's roadmap is that they are currently doing 85% of production at .21µ, with 15% of wafer starts at .18µ. Their goal is to be 100% at .18µ by the end of the year. I have not heard of any plans for 300mm by MU. I do not know when MU plans to begin production at 256mb, or if they plan to produce 128mb chips. I suspect it depends on pricing and estimated profit margins.
Larger wafers are a big cost savings especially when dealing with big dies. Imagine if you were putting one big square die on a round wafer - the amount of wasted silicon would be huge. On the other hand, the smaller the die, the better use you make of all the silicon, and the less cost savings for going to a larger die. Part of the reason that 300mm has taken off so slowly is the rapid transition to smaller geometries in the last several years which has kept the dies small, and has kept 300mm cost prohibitive. Eventually 300mm will be necessary, but I'm not sure if this will be at 256mb or 1gb. I lean towards the later.
From your article on Tiawan:
Winbond - currently making 64mb chips, geometry unknown, pilot runs of 128mb chips at .17µ in Q3 1999
Nan Ya - currently making 16mb chips at .28µ, moving to 2/3 64mb chips at .20µ and .175µ with the remaining 1/3 still at .28µ by the end of 1999
Winbond is clearly not far behind, but they are just doing the production work for Toshiba. They appear to be about 6-9 months behind MU, who is already doing pilot runs at .18µ, and have been doing so for some time. Six months is a good lead for MU, but not insurmountable.
Nan Ya looks like they are in deep trouble, if you ask me. They lost big bucks last year, and will almost certainly lose big bucks again this year now that memory prices have fallen again. They are making 16mb (buggy whips) at .28µ and selling them against MU who is making them at .21µ. Guess who can make them for less? Even by the end of the year they won't be where MU is today. It looks to me like Nan Ya should suffer the fate that bears have long been forecasting for MU. Bears, take an honest look at Nan Ya, and pretend for a moment that they are MU, and just think what you'd say about them. 2 generations behind on chips (Samsung is at 256mb, Nan Ya is at 16mb), 2-3 generations behind on geometries (They are at .28µ, the leaders are working on .18µ). Heck, Nan Ya even makes TI's DRAM production look state of the art. <G>
... they are fore an expansive surprise ---- Excuse me????
Carl |
