Not forgetting proliant servers, arguably Compaq's greatest asset
The saga has been gone over ad nauseum in the dec groups, ever since the cancellation, but suffice to say Alpha development costs were miniscule compared to itanium, performance was far higher and they were making significant profit. Hp were and are in bed with intel with itanium and at the time and itanium performance was underwhelming to say the least. Hp wanted itanium for a pa risc replacement and had sunk considerable time and effort into the project. Intel thought that it would be the next Big Thing that would monopolise the cpu market, with the added advantage that it was so complex that it could never be copied, unlike x86.
As things stand now, only a few vendors build Itanic (as it's known :-) systems and it's just another niche product, will never cover cost of development and whose performance has been overtaken by advances in simpler x86 technology, primarily via 64 bit extensions. It may seem smug, but most people who are interested in the advancement of the state of the art in computing find it quite amusing that intel and hp have fallen flat on their face on this, because the politics and nih factor have been plain to see all along to anyone with even half a clue. After all the money that had been spent and hype about performance advantage, no one involved could admit that the approach was wrong to start with, so they just kept and even now, keep going.
The latest itanium has only just caught up with Alpha in performance terms and that's after a 6 or more year interval and billions (No, not a typo) of $ spent on development. Compare that to the few 100's of million *overall* spent on Alpha development :-)...
Regards,
Chris