Itanium is dead and as most suspected, the end came with a whimper, not a bang a mere two months after the current generation was launched. Call it a mercy killing, call it long overdue, call it not worth caring about, but in the end, Oracle’s boot did the deed and no one cared.
A bit over a week ago, Intel finally announced what everyone knew was coming, Kittson, the 22nm successor to the 32nm Poulson would actually be 32nm. Instead of a shrink, Kittson is not a shrink any more, in fact it isn’t really much of anything. It was never meant to be a new architecture, and now has been reduced to essentially low hanging fruit and bug fixes if Intel even bothered doing anything at all. It uses the same socket and boards as the current Tukwilla and Poulson CPUs, and converges on a common socket with x86 Xeons.
If this sounds like an exciting product, well you are misunderstanding something fundamental, there is nothing worth a damn here even if you are locked in to Itanium hardware for some reason. This groundbreaking announcement was done in the most peculiar way, it was put out in a one paragraph note buried on Intel’s site with no fanfare, no press briefs, and no nothing. It is almost as if they didn’t want people to notice, or maybe they don’t care anymore either.
Why did Intel backpedal so hard and turn full shrink in to a barely warmed over B-stepping? These are multi-thousand dollar mainframe-ish CPUs, cost is not high on the list of concerns to customers buying the Itanium computers, system price, software licenses, and maintenance costs dwarf what Intel asks for silicon. So why wouldn’t Intel do what they promised with Kittson? Money, or lack of it. This is not to say that the selling price was unfair or untenable, but Itanium development costs have been subsidized by HP for years.
Lets face it, Itanium is made by Intel, but there is only one real customer, HP. While there are technically other companies making big iron with these beasts, their combined volumes are close to a rounding error compared to HP’s numbers. While HP sells the overwhelming majority of Itanium CPUs, they don’t actually sell that many, the market for this class of machines is small and shrinking. But it is lucrative, very very lucrative per unit, as long as you don’t consider one time costs. HP makes massive margins and profit off of the big iron category, and software with bundled support contracts are even more bottom line friendly. Just like IBM effectively funds fabs and Power CPU development with software and maintenance contracts, Itanium hardware is the gateway drug for HP’s enterprise business.
Given this massive profit margin, HP would likely do anything needed to keep Itanium alive and their customers hooked on this silicon smack. And that is just what HP did, they essentially funded Itanium development, Intel just carried it out. The price charged per CPU was not really a rational thing, no company could sell them for the prices Intel charged if they had to account for development costs. As long as HP made it worth their while, and they did, why shouldn’t Intel keep developing it? So HP kept paying, Intel kept making, and in spite of the fact that the chips were painfully awful, large corporate customers kept buying. They had no choice, once your enterprise software stack is dependent on something, it is far more money and pain to switch than to just pay up. If the product works OK, you have the enterprise IT version of detente.
But HP stopped funding Itanium development for some reason, and so Intel dutifully stopped development, both quite suddenly. Oh sorry, Intel decided to put out something with the same name as an expected 22nm shrink and update, but it is really the silicon version of whiteout over the name tag. And they announced it with a one paragraph tidbit on a dark corner of their site. Make no mistake though, the money flow stopped, and so did development, HP pulled the plug.
Why would they do such a silly thing? Why would they jeopardize their agonizingly lucrative enterprise customer lock in along with all the mostly profit service contracts that those entail? Easy, no customers. Why would locked in customers flee this lock in? Didn’t you just say that switching was more costly and more pain than sticking around? If they all left in short order, there had to be a reason, right? And there was.
In the enterprise software version of a death sentence, Oracle pitched a little hissy fit, pulled support for Itanium, and poked a stick in HP’s eye in the process. HP was a little annoyed by this direct attack on their most lucrative profit center, and blamed their ex-CEO Mark Hurd for the slight. Well blame and slight are a bit understating what happened at that point, multiple multi-billion dollar lawsuits flew back and forth, and surprisingly the dust settled very quickly. To the surprise of no one, Oracle lost the suits badly, painfully, and publicly, and there were a lot of zeros attached, but that part hasn’t been fully settled yet. Oracle was forced by the court abide by their contract with HP, meaning they effectively had to resume support and continue development for Itanium. HP proclaimed victory.
Unfortulately for HP, they won the court battle, but lost the war. As we said at the time, HP trounced Oracle in a court battle, but Oracle decisively won the war. If you are using enterprise software that is literally the lifeblood of your mission critical systems, and doing it on multi-million dollar hardware, software, and lots of custom code, you take uptime seriously. Spats between vendors are not acceptable, and the sheer number of zeros involved in anything sold to this space has traditionally meant that it is worth more than enough to all parties to grin and bear problem than it is to fight. Fighting means lost customers, and that means lost revenue with obscene profit margins, so no one really fights. Until the HP/Oracle spat.
Oracle forced customers to take sides, their databases, ERP offerings, and all the rest of the backbone of most large enterprises or HP hardware, pick one. Fortune 500 IT departments were faced with moving off their Oracle software or HP hardware, but which brought more cost and pain? HP used the courts to force Oracle to stop this “Us or them” offer, and the courts did exactly that. Technically. And as we said, no sane enterprise is going to trust that Oracle support from now on, would you rest the uptime of your mission critical systems on a court ordered support network between two vendors that hate each other to the point of mocking press releases?
Sure enough, every HP customer saw the writing on the wall just like SemiAccurate predicted. The court order meant nothing, Oracle said, “choose” in a way that meant it, and customers did. As we said in the earlier articles, Oracle knew where to hit HP to inflict maximum damage. When they fired away, the first volley sank the HMS HP Enterprise hardware in one shot too. As we predicted, the court ordered lifeboats were woefully inadequate, and the good ship is almost fully under water. Captain Kittson is putting on a comfortable set of sweatpants in preparation to go down with the ship, and they haven’t even shrunk a bit since they were last put on in November by Captain Poulson.
Oracle won the war, decisively, court order or no. This isn’t to say that Oracle will do anything but abide by the letter of the order, and do so with a genuine smile on their face too. The gutting of the Kittson shrink to 22nm says that HP agreed that it is game over and decided it was foolish to waste money on the shrink at this point. Intel did what they were paid to do on Kittson, essentially nothing, while pretending this was the intended path all along. To signify the death of Itanium, the world got a one paragraph statement, the line no longer justified even a press release. The one thing that now unquestionable is Oracle, HP, and Intel agree that Itanium is completely beyond hope at this point. About time.S|A
UPDATE: HP contacted us with the following statement as of 1pm February 14th, 2013.
On January 31, 2013, Intel posted a statement on intel.com, providing an update around its plans for "Kittson," the codename for Intel's next generation Itanium processor. HP and Intel have a long history of working together on Itanium processors and Integrity systems to deliver mission-critical infrastructure for the world's most demanding applications.
HP is committed to the Integrity product line, including ongoing innovation and development of a new line of Kittson-based Integrity systems in the future, working closely with Intel. The recent statement by Intel has no impact on those plans or on HP's ongoing commitment to our mission-critical customers.
Latest posts by Charlie Demerjian (see all)
- SemiAccurate has Skylake-SP die shots, sizes, and more - Jul 21, 2017
- Skylake-SP has a diverged core - Jul 19, 2017
- Intel’s Purley platform architecture is a step forward - Jul 13, 2017
- Intel launches Purley aka metal Xeons - Jul 11, 2017
- AMD’s Epyc has lots of connections - Jun 26, 2017