Profitability is a lower baseline. The real metric is opportunity cost. What else could have been done with an investment that took a decade to break even?
I agree opportunity cost is a better metric, but there's yet more to it. In another 10 years, Itanium might surpass x86 as the old architecture hits the wall and the new continues to speed up (if in fact it does).
It's almost a matter of insurance -- Intel can't afford to be stuck at ~3.1416 GHz :-) while competing architectures continue to improve. If in fact they can.
