Intel wasted a crapload of money on Itanium
Itanium has certainly not been a success in the general market, but the reality is that it was never supposed to be (although Intel was awfully hopeful on that end, with the idea being that a smaller subset of the Itanium core might eventually replace x86; that's not happening).
In the high-end server market Itanium has been moderately successful. Certainly HP isn't screaming bloody murder, and they've long since shelved their PA-RISC systems and are roughly competitive with the IBM (Power) and Sun (HyperSparc) offerings.
People keep forgetting that Itanium wasn't supposed to outright replace x86 on the desktop -- it was supposed to give Intel a foot in the door in the mid-end server and high-end workstation market. The workstation market has pretty much disappeared (the x86 market ate it alive; so Intel isn't suffering there), and the Itanium is giving them the entrance into the mid-range server market they wanted.
Although I'm still glad I didn't take that job in the Itanium software development group that I was offered way back when. It certainly didn't succeed as much as Intel had hoped it would, and most of that group wound up looking for new jobs.
Oh, and if AMD's large customer support is as abysmal as Caelistis says, then that's a good reason not to go with them. My company is almost certainly a large customer, but I'm not in that department.