InvestorsHub Logo
icon url

I_banker

08/19/03 10:34 PM

#11533 RE: SemiconEng #11531

Semi, the only piece of the analysis that was left out (and its a big piece), is the increase in the number of chips produced per wafer with the smaller geometry process. This presumably represents a great deal of incremental revenue.

Question is will the extra revenue offset all the headaches you outlined? Is it in fact like buying a new fab on the cheap?
icon url

sgolds

08/20/03 11:40 AM

#11590 RE: SemiconEng #11531

SemiconEng, tooling: Fab 25 is populated with 180nm equipment which used to be used to make processors. Thus their memory manufacturing is already based on last-generation processor generation, so the problems are not as onerous as you assert.

Now, AMD does not use a 'copy exact' since it has a single microprocessor fab, but Intel does not 'copy exact' from processor to memory (so I'm not sure how this point from your post affects anything).

Certainly there are differences, even when the memory processes are based on processor processes. Since migrating processor to memory technology is the fundamental behind AMD manufacturing, I suspect it is a lot quicker (and less problems) than starting from scratch!