I recall opening up the Itanium manual, and by the end of the architecture description, just despairing of the thought of trying to write a compiler for it. Itanium, I think, was ultimately a victim of its weirdness: it's too weird to really comfortably write assembly by hand; the compilers weren't really capable with its weirdness, so "regular" code was worse off than you'd normally expect. Raymond Chen has pointed out in several articles how the hardware took advantage of C's UB to do some really weird things--and this is an era where most developers expected UB to really be just implementation-defined behavior.

Combine that with the fact that the hardware development process seems to have been compromised from the start (if you told me the hardware architects never looked at anything other than 30-instruction traces of BLAS kernels, I'd believe you), and the insane hype that was built up for it... it's not surprising that it had an extremely underwhelming launch.