Intel defends PC, goes all-in on ULV, and speeds up Moore's Law

[nsane.forums]

Intel defends PC, goes all-in on ULV, and speeds up Moore's Law

At an annual investor relations event today in San Francisco, Intel execs took to the stage to defend the PC against the claims that "tablets" (currently just a euphemism for "iPads") will do it in, and to make a comprehensive case that the future belongs not to ARM but to Intel.

Of course, what Intel means by "PC" is much thinner, lighter, and more appliance-like than the Windows-based desktops and servers that the term has historically brought to mind. Indeed, the company was up-front about its attempt to redefine the PC in a direction that's essentially embodied by Apple's current MacBook Air—a thin, power-sipping, tablet/laptop mash-up that may or may not have a copy of Windows installed somewhere on it.

Intel's demo rooms and stage were filled with slender prototypes. Gone were the painfully bulky MIDs of previous years, and in their place were a suite of relatively attractive tablet and laptop prototypes that one hopes will at some point deliver on Intel's promise of a ten-hour battery life.

All of this change is part of Intel's plan to "reinvent and re-energize the PC," as Intel CEO Paul Otellini put it in his opening keynote. But the reinvention isn't limited to just the form factor.

ULV: the new normal

Otellini also announced that Intel is making a fundamental shift in the way that the company thinks about and approaches the design of its notebook chips. Until now, the bulk of Intel's notebook chips are design to draw around 35 watts of power—many of its notebook parts are lower, and some are higher, but 35 watts is the center point for Intel's portable lines. Going forward, however, the new center point will be in the 10 to 15 watt range.

To understand what a radical shift this is, consider the fact that the newly announced Core i5 and i7 ULV (ultra-low voltage) chips have a TDP of 17W. So, what Otellini is saying with this announcement is that, instead of being an exotic, expensive, and late-coming niche in Intel's mobile lineup, ULV processors are the new normal.

Under this new regime, Intel's SoCs will scale from about 500 milliwatts up to 10 watts, and the notebook chips will scale up from there.

The tablet Who Must Not Be Named

The iPad's success is clearly a sore spot for Intel, no matter how much the company protests that it isn't. The presenters' slide decks contained a number of images of tablets at various points, not one of which was an iPad. Throughout the day, the Apple's tablet was thunderingly absent from all of the tablet and portable talk.

In fact, at one point, Otellini put up a slide showing the growth of "smart devices," by which he meant only smartphones and laptops—no tablets or smart TVs. He justified leaving out the iPad tablet because PC and phone growth allegedly dominates tablet growth to such a degree that there was no point in putting it in.

Then there was another telling slide from Otellini, which showed laptop Internet traffic (in MB/month) outgrowing smartphone and tablet traffic by 3X. The conclusion that Otellini drew from this data was that laptops are growing much faster than smartphones and tablets. But anyone who has used a mobile data plan on both a tablet and a laptop knows that, for whatever mix of reasons, laptops are much bigger data hogs than tablets over the course of normal use. In other words, all that the slide really showed was that laptop users use a lot more data than smartphone and tablet users.

The reason that the iPad gives Intel heartburn is twofold. First, there's the fact that iOS doesn't run on x86, and probably never will. But the ARM-only iPad is also the one successful example of a tablet that anyone has seen so far, and that success has been driven by Apple's (apparently unique) talent for delivering an overwhelmingly compelling experience for end-users and, to a lesser extent, developers. So Otellini knows that his tablets will always have to compete with Apple's, and that can't be a happy prospect.

The second, and closely related, challenge that Intel's tablet ambitions face is the sorry state of the tablet competition. When Otellini mentioned tablet OSes, he dropped the names Android, Meego, and Windows quite a bit. Not one of these is even close to being a compelling iOS alternative at the moment, and Otellini's lack of excitement for any of them was evidenced by the fact that he didn't bother trying to show even one of them off. In fact, on the Meego front, Otellini made a few painful jokes and references to Nokia's very public rejection of the joint Intel/Nokia effort, all of which drew sympathetic laughter from the audience.

The upshot of all of this is that Intel's tablet is a horse without a rider. It could be a prize-winning thoroughbred, but all of the potential jockeys are so bloated and generally unfit for the saddle that it's not at all obvious how Intel's entrant can truly compete.

Countering the iPad with a laptop

Intel's answer to the tablet dilemma comes in three parts. First, Otellini declared up-front that real innovation in tablet computing has yet to fully kick in. "The tablet race is nowhere near finished," he said. "There's going to be a tremendous amount of experimentation here for several years." So, at some point, the argument went, someone besides Apple will make a tablet worth buying. For our part, we're hopefully optimistic that this is indeed the case.

Intel's second part is that the tablet doesn't so much replace the laptop as it augments it. One of the Intel presenters argued that a tablet is really a second or third or fourth device, and not a replacement for the PC. The strategy here was to pitch the tablet as a toy for the affluent; the same presenter claimed that half of iPad buyers in mature markets have income levels of $100,000 a year or over. He also pointed to the fact that, this past quarter, the MacBook line actually outgrew the iPad.

The third answer to the cannibalization story is that, even if tablets were to cannibalize up to a third of PC growth, it won't knock more than a few percentage points off the PC market's compound annual growth rate (CAGR). In other words, the PC market is so much larger and is growing so fast that even significant cannibalization by the still-nascent tablet market won't hurt it more than a bit.

Smartphone and Atom update

Intel didn't have any major smartphone news today. The company merely reiterated its plans to have the first Intel-powered smartphones hit the market in the early part of 2012.

Otellini referenced the Nokia breakup and said that Intel is shopping the reference design that it made with the phonemaker around to other OEMs.

Intel also took a moment to reiterate that its 32nm, 22nm, and 14nm versions of Atom will all be out over the next three years, with 22nm dropping next year and 14nm the year after that. This represents an acceleration of Intel's process roadmap—the chipmaker is essentially stepping on the gas to get Atom iterations out faster, cycling through the tick-tock cadence at a quicker rate.

Intel also put its upcoming 22nm Atom core, codenamed Silvermont, on a slide for the first time. Silvermont is speculated to be the first out-of-order Atom design. This is probably true, and it will give the part a significant boost in performance vs. previous-generation Atoms.

Following Silvermont at 14nm is Airmont. Not much is known about Airmont beyond the fact that it's due in 2014.

View: Original Article