When I read the original article doesn't it say anything about a Xserver Ultra to be presented during the MacWorld. What it say is this: "According to our sources, Apple might be launching a series of Xserve "Ultra" powered by Itanium 2 in a near future."
That is something rahter plausible, isn't it.
No, it's not plausible. "Possible", yes. "Plausible", no. If Apple had gone with Itanium
instead of Core Duo on the original Intel switch, maybe. But Itanium is a completely different instruction set than Core Duo/Core 2 Duo/Xeon. (And Intel has never even made an attempt to make Itanium-family chips low enough power for mobile computers. The "lowest end" they ever even attempted to market it was for high-end workstations. They stopped marketing it as a workstation chip a few years ago, now it is solely a server chip.)
The chips used in everyday PCs, and now in Macs, are "Intel x86" instruction set computers (Intel calls it "IA32".) The latest ones support what is usually called "x86-64", (called "EM64T" or "Intel64" by Intel, and "AMD64" by AMD,) which are just a 64-bit extension to the old x86 instruction set architecture. x86-64 chips are 100% compatible with 32-bit x86. They have to be, since at their core, they are still x86. The actual instruction set is called a "CISC" instruction set, for the way the instruction set itself work; although the actual processors are, deep down, mostly "vaguely RISC", and have been for over a decade, doing internal conversion from the x86 instruction set to their own native "micro-ops". These "native" micro-ops are *NOT* available to programmers, because they can change wildly from one processor generation to the next, based on what Intel learns about how to improve performance.
Itanium on the other hand, is a completely different instruction set. Not
at all compatible with x86. It uses the "IA64" instruction set, which is a "VLIW" instruction set. It is so completely different, that when the very first Itanium was released, Intel embedded what was essentially a small Pentium Pro processor in it, for compatibility with x86. It turns out that software emulation (like the old SoftPC or Virtual PC for PowerPC-to-x86; or PearPC or Rosetta for x86-to-PowerPC,) can emulate the x86 instruction set so much better than the little embedded x86 chip, that Intel removed it from the Itanium 2.
This means that Apple would have to develop *ANOTHER* Rosetta-like software to allow an Itanium-based Mac to run Intel-Mac software; and yet another emulator to allow an Itanium-based Mac to run PowerPC software! This is in addition to the raw challenge of compiling OS X for Itanium; which would mark a FOURTH architecture: PowerPC, Intel x86, and ARM (iPhone/iPod touch.) Well, at least they don't have to support 680x0 any more...