Re: IMHO it's a fab war. What's AMD's fab capacity?
Not all FABs are the same. What's important is the right FAB at the right time. What matters in the coming year are sub .18 copper FABs.
IBM, Motorola, AMD, and TSMC have worked hard over the last several years so that they would all have them.
It's a FAB war all right, and Intel is coming into the battle unarmed. Estimates vary, but given that the cost of $120 processor and a $500 processor are about the same (at $30 to $60) it's easy to see that a disproportionately large fraction of the total profits in the microprocessor industry come from the 10% to 20% of chips that constitute the high end.
Intel isn't going to be particularly competitive in that area during the coming year.
In large systems, SUN is vulnerable, but IBM is most likely to take any market share that SUN loses because IBM has the FAB capacity to produce fast, high end chips.
In small servers and high end desktops and workstations, AMD is very well positioned to take a significant fraction of a market that has been 99% Intel for the past decade. Because AMD has the the FAB capacity to produce for that market - and Intel doesn't.
If IBM (and perhaps Alpha/Compaq fabbed by IBM, if recent rumors are correct) keeps Intel from expanding into the large system space, and AMD takes half of the small server and high end PC/Workstation, it won't be good for Intel.
40 million processors at an ASP of $140 is not as good as 30 million processors at an ASP of $210. And it will be the lack of FABs that causes the change.
AMD broke ground on its .13 copper FAB in late 1996, Intel broke ground for its first .13 copper FAB a few months ago.
Regards,
Dan |
