To: A. A. LaFountain III who wrote (57170 ) 5/4/1999 4:29:00 PM From: Saturn V Read Replies (1) | Respond to of 1572405
Ref your comments " I believe the description of the fab was "eating like an elephant, but defecating like a canary." Hold it, that's not Jerry Sanders, that's Gordon Moore. And he wasn't describing AMD's Fab 25, he was talking about Chandler when it was ramping up. Having seen that happen to Intel, I believe that what AMD has been going through is NOT company-specific. Instead, it appears to be something that any IC vendor has to go through in order to get to sufficient volume to have multiple fabs allow some diversification of risk. Copy Exact is nifty and everything, but to implement it, you need multiple fabs. How do you get to multiple fabs? "" I am surprised that you remembered Gordon Moore's comment. He made the comment in 1983 after major problems in starting up the New Mexico Fab plant. To be more specific the New Mexico plant was the world's first 6 inch Wafer Fab. All the Wafer Fab Equipment was first of a kind and each had Serial No 1. So some technical problems were inevitable. The inherent technical problems were compounded by the Engineering Team. They were relatively new hires and still had the 'Cowboy Mentality ' , and ignored the Intel 'Copy Exactly' culture. They needed to make some changes to the process to port it to 6 inch wafers, but they used that as a license to change everything. It was a disaster which lowered Intel's profits significantly for a year. However Intel always learns from its mistakes and vowed never to have that happen again. Intel enforced the 'Copy Exactly' methodology more rigorously and has avoided leading the way to larger wafer size. Since 1983 Intel has honed its manufacturing skills significantly and leverages its economy of scale better than the industry. For example all the fabs which run a process are continuously sharing information and data, and have the identical equipment. This brings up new fabs relatively very smoothly. AMD has not been as succesful at leveraging economies of scale. Fab 25 took the 0.35 micron process from the Sunnyvale Design Center and made several changes. Thus in 1997 the K-6 start up was a disaster until the Engineers from Sunnyvale flew in , took control and the yields were restored. Furthermore the Dresden 0.18 micron process and the Texas 0.18 micron process are going to be very different. Thus AMD will not reap the economies of scale, as Intel will. AMD is doomed to have a higher cost of manufacture than Intel. Thus a "40 percent lower sales price than Intel" business model will continue to bring major losses.