Bill - Re: "...if 70 % of manufacturing capacity will .25 micron next year will the other 30 % be .18 micron ?"
No - this year and next year Intel will be converting several fabs from 0.35 to 0.25 micron processes. Hopefully, by the end of 1998, five fabs will be running 0.25 micron processes at some level of production - probably approaching 50 - 70% of Intel's x86 CPU production.
The 0.18 micron process is in development and won't see first "real" production until early 1999. This program may or may not be coupled to the 300 mm wafer process. I have heard several conflicting reports on this so I am not sure of Intel's current plans for this "harmonic convergence".
Re: "0.35 micron process"
I would expect Intel to maintain 0.35 micron production for older chips - primarliy chip sets, microcontrollers, support circuits, etc. Thus, Intel will be maintaining this process for quite awhile, but will probably cease x86/Pentium MMX/Pentium II by late 1998 or early 1999.
Re: "I assume they build in these countries to enter the areas particular markets ( also cheaper labor costs)"
Correct - "local" manufacturing is now critical to getting accepted in the local markets from a sales standpoint, now that "global" consumption of PCs and components is the norm. Formerly (not now), labor costs were the prime driving factor.
Re: "What is included in the QC process at these facilities -IS it time consuming -what is the rejecton rate ? "
This could take days to discuss. However, let me summarize it one statement - ZERO Re-Work!
In other words, Intel does no re-work - if the parts don't meet Intel standards, they are SCRAPPED.
The significance of this is quite apparent. Machine & Equipment QC, materials QC, and process QC all have to be rigidly maintained. Intel long ago gave up the notion that "Quality can be inspected in".
Thus, primary emphasis is on getting the right equipment for a process, training and understanding the maintaenance and set up of these machines, and dialing in the process to insure ALL PRODUCT meet minimum quality standards.
Long, long ago Intel thought 90% overall yields was difficult to attain. For the last decade or so, I think their asse mbly yields are closer to 98 - 99%. This is critical considering the cost and the potential lost sales revenue of products that would otherwise be be rejected.
As a further comment - Intel's Reliability and Quality Control departments have ALWAYS been independent of manufacturing and have exercised independent control over their domains. That is, these departments are NOT subordinate to manufacturing. They "dictate" what can and cannot be made and done in the production plants.
Paul |
