It's not exactly an obituary, but yesterday's story from CNET reporter Stephen Shankland sums up the 11 years of HP's Itanium project with Intel as "a cautionary tale." Yesterday we sat down with an old friend, Terry Floyd of 3000 ERP experts The Support Group, and we mentioned Itanium briefly. "Could it really be 11 years ago?" we both asked, noting that it seemed like a long time for a chip that hasn't set the world on fire.
As Shankland's story notes, HP and Intel were predicting a blazing future for their joint project way back then. First it was code-named Tahoe, then Merced. Finally the pair of technology giants got first silicon shipped, but the competition wasn't standing around waiting for the release. AMD, IBM and even Sun rolled out their 64-bit entries. Shankland notes that Digital's Alpha fell in the chip competition — but you might argue that the Alpha demise was more about Digital being acquired by Compaq, which was then acquired by the makers of Itanium.
Itanium has a future in the HP 3000 marketplace. Shankland's story reports that Gartner says four of every five Itanium servers shipped during the third quarter carried an HP nameplate. Itanium is the successor to PA-RISC — which is still outshipping Itanium, according to Gartner's figures. But back in the 1990s, CIOs and software companies were told to believe Intel would conquer all with the design that was born in HP's labs.
If there's a caution in HP's Itanium story, it might be this one: Shipping something late is never in line with dominating a market. Late chips don't run any faster when they arrive late. HP will prosper by its investment in Itanium, if you believe that PA-RISC is out of gas. Itanium is the only long-term platform for HP-UX, so that makes the chip's fortunes important to the 3000 community which is migrating to HP's Unix. Some said that HP did its deal with Intel 11 years ago to gain strength of numbers in massive market share. Itanium is unlikely to accomplish that goal, so HP is left with a secondary success: Getting to shut down its chip fabrication and design operations, so it can spend that R&D money someplace else.