Posted by: Peter Burrows on September 15, 2005
Itanium. Just a few years ago, it seemed certain the word would be as important a part of the techie lexicon as “X86” or “Windows”. Back then, it seemed Intel Corp. had more than enough market power to convince scores of computer companies, thousands of software developers and millions of server-buying customers to embrace its powerful new 64-bit chip family. And no company looked better positioned to capitalize than Hewlett-Packard, which decided in the late 1990s to contribute some of its most prized technology to help Intel develop the new family of chips. That way, HP could get out of the expensive business of making its own proprietary microprocessors and reap a nice royalty everytime an Itanium was sold—and get a big leg-up on building servers to make best use of the chips.
It didn't work out that way. Itanium not only hasn't taken off; it's barely gotten off the ground. There are lots of reasons. During the tech downturn, many customers opted to go with cheaper servers, often tied together in server farms using sophisticated grid technologies that let them handle tasks that had been done in the past by high-end screamers. What's more, AMD roared to life in the server market, with chips that had Itanium-like power, but which could also run x86-compatible code. After waiting years for lift-off, many computer makers have decided to bag it, instead. IBM backed away earlier this year, and today comes news that Dell will do the same.
So what about HP? In a recent interview, HP Labs head Dick Lampman made it very clear: HP is sticking to its guns. The company's story has changed a bit; rather than conquer all of server-dom, the Itanium will now be primarily used in high-end computers, as a way for HP to go after the $20 billion UNIX replacement market. "I never believed Itanium would replace x86. I just never felt it would," insists Lampman. And he says the Itanium is a shoe-in to become the standard in this market. One reason is because he's certain--unlike some others--that Intel remains fully committed to the chip. He notes, for example, that Intel is adding multi-threading capability to the chip. That's proof, he says, that Intel isn't winding down its development effort.
I asked whether it was even possible for HP to resurrect its own PA-RISC processor development effort at this point, if it decided to reverse course. Lampman didn't even bother to answer. "I don't know, and it doesn't matter, because it's not what we're going to do. We're doubling down on Itanium," said Lampman.
I wasn't able to reach Lampman today, to see if Dell's decision had changed his view. But a spokesperson says it won't--and points out that Dell, which focuses on selling lower-end volume servers, was never going to be a major driver of the Itanium market in any case.