Re: AMD has no server products, no workstation products, no notebook products, no new fabs and no .13u process...
Yet their sales and profits continue to increase relative to Intel. AMD has new server, workstation, and notebook products in the final stages of development.
AMD's rate of gain on Intel would accelerate if they entered these additional markets.
The biggest cloud on AMD's horizon, IMHO, is Tualatin. PIII was a good core, and if Intel manages to fix the scaling problem, Tualatin could become a tough competitor. But a strong Tualatin would destroy P4, Itanium, Foster, and McKinley - so there are some "political" as well as technical issues.
Northwood and McKinley are also coming, but both initial core implementations show signs of fundamental flaws - Itanium -> McKinly X86 performance is unacceptable, so it may as well be an Elbrus 2K as far as buyers who want to run any existing software are concerned. P4->Northwood has embarrassing IPC performance and hasn't been scaling nearly as quickly as it must to compensate for that problem. If those problems are fixable, unlike K6's and .18 PIII's scaling problems, it will present a challenge for AMD.
There are many clouds looming over Intel. Near term, the fact is that the old Athlon core, which looked for a while like it was stuck at 1.1GHZ on Copper, clearly is good for 1.3, and perhaps 1.4 or even 1.5GHZ.
The next cloud for Intel is the Palamino/Morgan core, which wouldn't have been too scary if it had added 25% to 1.1GHZ, but is downright terrifying if it's adding 25% to 1.4GHZ. It means that AMD ready for a 2GHZ P4 without going to .13.
The most important thing about this new core may be what it could do for Austin. If the new core and subsequent tweaks means Austin will be able to ship .18 1.33GHZ to 1.46GHZ Aluminum Durons through 2003, it means that AMD's low end chips could stay competitive with Intel's .13 offerings without AMD needing to build a new FAB.
Another cloud on Intel's horizon is the rumored near term release of low power Athlons for notebooks. We've already seen what happened to Intel ASPs when it started facing competition in desktops.
Another cloud on Intel's horizon is the rumored near term release of SMP Athlons for servers and workstations. We've already seen what happened to Intel ASPs when it started facing competition in desktops.
Another cloud on Intel's horizon is AMD's rolling conversion to .13 at its existing Copper FAB. AMD has far more Copper FAB experience than Intel, so there is a decent chance that this will be a relatively quick, relatively painless conversion for AMD. AMD certainly did a great job moving from .25 Aluminum to .18 Aluminum at Austin. Intel has to learn to do .13 at the same time it is learning to do Copper, which has got to complicate the process.
Another cloud on Intel's horizon is AMD's adoption of SOI technology. Intel claims that SOI won't provide a significant performance improvement, but they said the same thing about copper at .18 - and it sure appears that Copper is helping AMD at Dresden.
But the Thunderstorm headed Intel's way, of course, is the Hammer core, which lets AMD market a full speed, X86 compatible, 64 bit processor against Intel's incompatible 64 bit processor and 32 bit X86 processor.
AMD's next CPU FAB is scheduled for 2004 - the time frame for the 153nm stepper technology to enter production. Intel could well find it's fresh $12 Billion FAB investment obsoleted overnight. Just like AMD's Copper Dresden FAB obsoleted $6 Billion worth of Intel .18 Aluminum investment. Intel could head into 2005 with its cash exhausted and its FABs obsolete.
How's that for a cloud?
Dan |
