Re: From what I have seen of Intel's plans, driving AMD out of business (Chapter 7 is ideal, Chapter 11 acceptable) is the goal.
It's a tough battle, but we're just now coming to the end of Intel's period of advantage (first to .13) and entering a "tied" period (both are at .13 copper).
But, beyond that, AMD expects to roll out a new process, SOI, and a new architecture for desktops, notebooks, and server, X86-64.
Intel expects to ship a specialized, lower power chip, Banias, and is working hard to get to 0.09 before AMD can make SOI high volume. These days, the performance gating factor seems to be as much power consumption as raw chip potential, and P4 has hit 83watts average / 94watts peak at 3.06 - maybe they can cut power use with "another spin of the silicon" as Jerry used to say.
Moving to copper made a big difference for both companies, and AMD expects that moving to SOI will make just as much of a difference. Intel is hoping to use strained silicon to compete with SOI, but SOI has been a proven technology for years, strained silicon is bleeding edge new - maybe Intel has enough resources to pull it off.
If AMD's Athlon-64 and Opteron are anything like they're expected to be, and both chips are close enough to shipping that that looks like a pretty good bet, then Intel's Itanium has been pretty much limited to very low volume specialty use. AMD also has a big jump on 64-bit architectures, and by 2005 32-bit chips may well be limited to sub $500 machines - and Intel could be trapped in that market.
Intel has decided that the next generation (64-bits) will either be a proprietary Intel generation (Itanium) or somebody else's generation. That's a bet the company move.
We saw from what happened with Rambus that it takes more than a year for Intel to switch architectures, even for something as simple as which memory is hanging off the North Bridge. Changing their main instruction set architecture from Itanium to Yamhill would almost certainly take longer.
Two to three years from now, AMD hopes to be able to ship very high volumes of their 64-bit design from a 12" .09/.065 FAB. If Intel isn't ready with a well established, widely accepted, 64-bit architecture by then, Itanium or Yamhill, they're toast.
If Intel can keep AMD sufficiently resource starved so that they can't increase their FAB capacity, then Intel can afford to take their time making Itanium the standard for 64-bit chips and keep AMD marginalized.
It's going to be an interesting few years. |
