wbmw, that was not a caricature of you, but of the poster you suggested had an "interesting perspective".
Regarding Itanium lingering on for several years in a niche, and thereby coexisting with Opteron, sure, that's likely, but given the investment in, and goals for, Itanium, I think that qualifies as failure, don't you?
And on the need for giant, single-image mainframes, I suspect that there will be very few (if any) applications which will not be equally (if not better) served by clusters with ultra-high-speed interconnects. We'll see how this stuff evolves, I guess.
Doug