Intel's Silverthorn Draws Doubts

CEO Otellini is betting on a new power-sipping chip aimed at mobile Internet devices, but analysts are skeptical it will be a silver bullet

When Intel CEO Paul Otellini took the dais at a May 3 analysts meeting in New York, he discussed plans for a processor that will be aimed at everything from low-cost mobile PCs to handhelds. The skeptics in the Marriott (MAR) conference room might have been forgiven for thinking they had heard this message before.

Intel (INTC) has struggled for growth in a personal computer market that in areas of the world is reaching maturity, and at various stages of history the company has taken aim at new markets like wireless phones and bargain-basement PCs—too often, unsuccessfully. Case in point: the mobile-chip business sold to Marvell (MRVL) last year.

So it wasn't hard to feel a touch of déjà vu when Otellini spoke about Silverthorn, a tiny, power-sipping chip that will be sold to companies making mobile Internet devices and stripped-down personal computers that would be as portable as a mobile phone, but as functional as a desktop PC.

Less Power, Less Cost

This time will be different, Intel says. The mobile devices will really be PCs—smaller maybe, but just as powerful from a computing standpoint, more or less, as the PC or notebook you're using now. Otellini described Silverthorn as being intended for devices that are "optimized for cost, power, and portability," sitting somewhere between today's notebook PC and a smartphone, like Research In Motion's (RIMM) BlackBerry or Palm's (PALM) Treo. "Think of this as the evolution of the cell phone from one perspective, but also the natural evolution of the notebook from another," Otellini said. "It's a full Internet-capable device."

Already, Intel sells chips designed to consume little power. Silverthorn, launching next year, will be one-seventh the size and consume one-tenth the power of a conventional processor chip on the market today. And when combined with other chips in a chipset, it will require one-quarter the space and consume one-quarter the power. Holding up a silicon wafer with 2,500 Silverthorn chips etched on it, Otellini said the part will cost less to produce than any Intel product since its 286 chip, produced from 1982 to 1986. "Think of all the places we can put this product; think about the margins and the costs that we can bring to bear on computing," he said.

Combine it with 3G wireless or WiMAX, and you'll really have something that consumers and businesses will clamor for, he hopes. Otellini says mobile Internet devices could be worth $10 billion in sales by 2011.

Restructuring Under Way

Otellini held up a prototype device that looked somewhat like an oversize cell phone, running Ubuntu, a version of the Linux operating system, though he said such devices could easily run full versions of Windows. It all sounded very interesting until he used the phrase "system on a chip." That's technospeak for an entire computer—processor, graphics, chipset, and all the rest but the memory—all crammed onto a single piece of silicon. And it can't help but revive memories of another Intel-sponsored effort to build a system on a chip. Code-named Timna, it was originally intended for release in mid-2000 and was to be aimed at budget PCs, but it was summarily killed before it ever saw the light of day.

Silverthorn wasn't the only matter for discussion. Intel also sought to portray itself as well on its way to completing a broad restructuring it announced at an analysts meeting in 2006 (see, 4/28/06, "Intel on the Offensive"). Otellini said the company is still on target to reduce $3 billion in costs by the end of 2008 and is making progress on ramping its next major manufacturing technology advance—chips built with design elements at 45 nanometers in size, on which the Silverthorn chip will be based.

But Silverthorn was one of the main selling points, and some analysts weren't buying it. "It makes sense to go after new markets," says Edwin Mok of Needham & Co. in San Francisco. "But it's really hard to speculate how this market will develop in the next few years. It might be a really hot area in two years, or it might be nonexistent."

Learning From Failure

While acknowledging the apparent similarity between Silverthorn and previous chips aimed at handhelds, Sean Maloney, Intel's chief sales and marketing officer, accentuated the differences. "Skepticism is healthy," he says. "The idea is to shrink a PC onto a chip, and that enables you to go after a very low cost. And there are plenty of people in the world who will buy computers if they are cheaper and smaller."

Previous efforts such as Intel's XScale chip for handhelds were based not around its main chip technology for PCs, but around a core technology called ARM licensed from ARM Holdings (ARMHY) in Britain, while Timna wasn't aimed at exactly the same low-cost segments that Silverthorn is. "It has full PC compatibility," Maloney says. It will enable "full PC functionality in a handheld device or just a cheap computer which has full Web-browsing capability."

And to be fair, Intel never fails completely, it just learns from those failures, says Dadi Perlmutter, senior vice-president of Intel's Mobility Group. He was there when Timna was in the pipeline and also when it was killed. "I view Timna not so much as dead, but that its heart was transplanted into other things, and we kept some of the genes," he said. "The key learning from Timna was that you could build a low-cost and low-power core and just apply it to different market segments." Some of those lessons learned with Timna were applied to Banias, code name for the first version of Intel's Centrino platform for notebooks, first released in 2003. "Some of the benefits that we learned there would also be applied to this one," Perlmutter said.

Demand for New Devices?

Even so, the market Intel aims for doesn't exist yet, though interest is beginning to grow, albeit sparingly, around ultra-mobile PCs, for which Silverthorn would be suitable. Cool and innovative as the concepts connected to Silverthorn may be, a new market would have to grow out of nowhere over the course of three or four years to get anywhere near Otellini's stated goals.

And there are plenty of hurdles. One is selling consumers on yet another digital device to carry around, says Ashok Kumar, an analyst at CRT Capital Group. "The key question is will there be a consumer for a new device somewhere between a handset or PDA and a notebook," Kumar says. "I'm not sure how Intel would work out the business model here. But Intel will have consumer fatigue to contend with. Do people really want to carry another device? They won't unless this is some kind of quantum improvement over what's available now."

Before it's here, it's on the Bloomberg Terminal.