Intel says it has the upper hand in ongoing fight with physics

d1x_graves.JPG

Intel is spending $6 billion -- or more -- to build its massive, 2.2-million-square-foot D1X research factory in Hillsboro, where scientists will develop new ways to continue shrinking computer chip circuitry. Intel researchers are currently developing the company's next-generation, 10-nanometer microprocessor inside.

(Mark Graves/The Oregonian)

SANTA CLARA, California – Intel manufacturing chief Bill Holt showed off his "baby pictures" at the company's investor meeting here Thursday.

The fuzzy, black-and-white images weren't of a cuddly infant – instead, Holt showed off electron-microscope photos of the tiny features atop the chipmaker's newest transistors.

One of Bill Holt's "baby pictures" an electron microscope image that shows Intel's latest, 14-nanometer processor. The "fins" that rise up in the picture help reduce power usage while enabling faster processing. They're taller, thinner -- and more fragile -- than prior versions.

"People are very proud of these pictures," Holt grinned.

Indeed, chip architecture has always been Intel's pride and joy. The company is spending $6 billion or more on its new D1X research factory in Hillsboro to craft future generations of chips.

Those tiny "fins" atop the transistors on Intel's new, 14-nanometer chips enable smaller, faster, more energy-efficient microprocessors – giving Intel-based computers and other gadgets a performance advantage that it says rivals won't match for years.

As with children, though, some things are outside the parent's control. In this case, Intel is continually bumping up against the laws of physics.

Features on its transistors are just a few atoms across, and those new fins are both taller and – evidently – more fragile than prior generations of chip architecture.

"Making those kinds of changes at these small dimensions is actually very difficult," Holt said.

The 14nm processor is arriving months behind schedule and Intel told investors Thursday that it's still behind the curve on reducing defect rates on the new chips, a costly setback since defective chips cannot be sold. The new chips are more expensive to manufacture, too, because tinier features require additional steps to imprint the circuitry.

In an interview with The Oregonian last month, Intel chief executive Brian Krzanich said he thought the troubles Intel has had with 14nm "are not going to be unusual" with future generations of chips featuring even smaller circuitry. And yet Intel said Thursday the cost savings from new chips are handily outpacing the additional expense required to make them.

Intel's last processor, its 22-nanometer chip, had 960 million transistors. The new, tinier design packs in 1.3 billion transistors on a chip that's 37 percent smaller – reducing production costs by putting more chips on each silicon wafer.

And while Intel closely guards details of its manufacturing process, Holt said Thursday he is "very confident" those cost savings will actually increase in the next two generations of Intel's chip technology, at least four years in the future.

"It will be below the historical trend," he promised investors. "That we know."

The steady reduction in circuit sizes is known as "Moore's Law," named for Intel co-founder Gordon Moore, who predicted the transistors on chips would double every two years in a 1965 academic paper. It's been Intel's guiding maxim throughout the company's history, and Intel celebrates the 50th anniversary of Moore's Law next year.

The future of Moore's Law is in the hands of Intel researchers in Hillsboro, who work in secret at the company's Ronler Acres manufacturing campus to devise new ways to advance chip technology. Intel said Thursday that it will be even less forthcoming than usual about its next-generation, 10nm chip technology than it has been in the past because competitors are using the information to plot their own strategy.

Intel is a great manufacturing company, said Kevin Krewell, principal analyst for Tirias Research, but sometimes it doesn't think hard enough about the ways its technological advances can be used. Intel has admitted it was caught flat-footed by the rise of tablets and smartphones, whose manufacturers have favored customizable chip designs from ARM Holdings over Intel's processors.

"That manufacturing-driven culture doesn't lend itself to being more creative on the product side," Krewell said.

Intel is counting on its leading-edge production to help it win over mobile device makers. But Krewell warned that contract chip manufacturers, which are called foundries, offer cheaper alternatives to make those chips. And that can outweigh whatever performance improvements Intel promises.

"Even though the foundries may not be at the same point as Intel in terms of manufacturing technology," Krewell said, "from a cost point of view they can actually do pretty well."

If you're betting the company, though, Christopher Rolland of FBR Capital Markets told clients Friday, Moore's Law "presents as good a business plan as any in the technology industry."

Sure, the PC market has struggled and Intel has stumbled in mobile technology. But Rolland wrote that Thursday's meeting demonstrates that the world's largest chipmaker's research gives it a long-term advantage few others have.

After Intel shares rose 4.6 percent Thursday to $35.95 – their highest close since 2001 – Rolland raised his price target on the company to $40, from $36.

"We are increasingly confident that Intel can opportunistically extract value from the extra transistors afforded to it through the best silicon manufacturing operations in the world," he wrote.

-- Mike Rogoway; twitter: @rogoway; 503-294-7699

If you purchase a product or register for an account through a link on our site, we may receive compensation. By using this site, you consent to our User Agreement and agree that your clicks, interactions, and personal information may be collected, recorded, and/or stored by us and social media and other third-party partners in accordance with our Privacy Policy.