Intel wants caviar, not cod roe, from IA-64
Peregrinations of a roving Vulture
Column A stretch of water was all that separated the King of Sweden and the King of the Itanic, Craig Barrett last week. And, we were assured by an earnest cab driver in Stockholm, you can still catch big fat salmon in these waters -- a fact certain to delight Intel's "Ice Man", a lover of outdoor pursuits.
The fat salmon which Intel wants to land over the next three years, however, come in the somewhat protean shape of large corporations, deploying the firm's IA-64 marchitecture and turning those horrible UltraSparc, Power PC and Compaq Alpha sharks into nothing more than prickly sticklebacks.
We know, from an Asian Pacific roadmap, details of which we released earlier this year, that Intel's pricing for its Itanium will become available to its OEMs in the middle of July. But that doesn't mean that Intel will launch the Itanium in July, a fact Barrett confirmed during a press conference he held after his keynote speech in Stockholm.
First, he said that Itanium chips were in production, rather than just samples. "The launch probably won't be in July, and it won't be in December," he said. "We are planning to follow Merced (the former codename for the Itanic) aggressively with McKinley a year later. "You'll see early adopters [of the Itanic] and then you'll see other adopters later."
Remember that the Itanic will be produced on a .25 micron process, and that McKinley, Intel has said, will become available towards the end of next year and will use .13 micron technology. Remember, also, that before Barrett took a plane to Stockholm, he cut the ribbon and announced a $2 billion expansion of his firm's Leixlip, Ireland fab. That, like the fab in Albuquerque, New Mexico, will move to .13 micron technology.
This poses one of many questions for Intel on its IA-64 strategy, and incidentally, for Barrett, who executed the firm's successful factory plans before he took over day-to-day running of the CEO job from Andy Grove.
First, we already know that Intel plans to move its production to .13 micron production, not just on its IA-64 family, but on its Willamette 479 chip, just as soon as it possibly can. This, as Intel's supply travails with its CuMine processors over the last nine months, is not simple. It now has five, maybe six fabs producing at .18 micron but has already vowed to shift to .13 micron. How long does it take to produce a .13 fab? Rather more than 12 months, we'd warrant. That would suggest that Intel, realistically, has no real hope that its IA-64 bricks will start appearing in volume until 2002 at the earliest. In the meantime, it can supply limited quantities of .13 McKinleys from the development plants it already has in place by this time next year. How many Willamette 479s can it deliver from its .13 fabs during Q2 2001?
Secondly, lawks-a-mercy, Intel is dependent on the success of its software partners in porting the necessary doobries to its IA-64 platform. As we suggested some days back, the Monterey alliance appears to be on target for so-doing, the Linux crew are also doing pretty well, but it appears that Microsoft Win64 development has a way to go. It wouldn't be the first time, by any means, that Microsoft has held Intel's plans back...
Strangely, even though Chipzilla's Dr Albert "even my mother has heard of copper interconnects" Yu, said two years back that his company has booked the Palm Springs Conference Centre for the next five years "because it was cheap", the next Intel Developer Forum will occur, in late August, in San Jose. Next February, it will be back in Palm Springs, we understand. Would it be beyond the realms of possibility that San Jose has been chosen to be the springboard for the launch of the IA-64?
One hack asked Barrett a question about how much Intel's chip biz will contribute to the firm's bottom line in the next five years, and the answer was illuminating.
Barrett said: "The majority of our revenues come from the IA (Intel Architecture) group. I would anticipate as you go forward to 2005, our semiconductor and chipset business will grow, but not as fast as the rest of the company. Five years from today, the majority of our income will still come from semiconductors".
Nevertheless, this means that Intel has to walk on the edge of a razor as it implements its IA-64 processor strategy and its shift to .13 technology and copper over the next eighteen months to two years. It's not frightened of walking on razors and, unlike Microsoft, every time it takes a step forward it has to dig into its deep pockets to the tune of $2 billion a fab, and trust in its roadmap and its undoubted expertise.
As AMD's Jerry Sanders III said so memorably some years back: "Only real men have fabs". If Sanders was right, and fabs equal testosterone, Barrett's Chipzilla is still giant to his Chimpzilla.
The rest of Intel's 80,000 strong workforce, and the INTC shareholders, must be hoping that Barrett -- the "Ice Man's Ice Man" -- will have the iron determination will help him to land the monster IA-64 sturgeon, keep the caviar on the table and hold the wolves at bay. ®