HT is an open standard run by the HT consortium or some organization right? Too bad Intel is too ego bound to not just use HT, more open standards would be good.

Folks, this has nothing to do with instruction sets. This is about chipsets. Intel is desperately trying to keep Itanium alive: it's actually reasonably performance competitive in its niche, but it's not cost-competitive... particularly with all the recent FP improvements in the x86 chips. By creating a common chipset for both Xeon and Itanium, they can bring the economies of scale that x86 enjoys at the platform level to Itanium. So the overall system cost of Itanium comes down somewhat (of course they still have to keep pushing the Itanium CPU itself forward to maintain performance leadership, and that's something they've had trouble with in the past). There may be some opportunity to pioneer new technologies that will filter down to desktops, but this is primarily aimed at servers and it's all about how the CPUs talk to each other and to the rest of the system. This is not about running x86 code on Itanium, or mixing instructions, or anything else.

At this point, do the architectural innovations of Itanium even come close to challenging the sheer momentum of the established x86 (and x86-64) base? Intel should know better than anyone that market momentum trumps technical superiority -- this principle is what has gotten them to the dominant position they enjoy today.

Does it mean that IA-64 and x86 instructions will eventually be decoded into the same kind of microops on the fly and both designs will virtually merge?

Hehe, everytime I see the words CSI, I think NIH-syndrome.

Those plans for common platform with the CSI aren't exactly new (and the zdnet interview isn't really about that). They have been known for 2 years now (http://www.techbuilder.org/news/60404730).

Must find whitepaper!