Jump to content

Intel's Army of Chip Makers Fights for Dominance


jamotto
 Share

Recommended Posts

Link

quote:

By Daniel Sorid

CHANDLER, Ariz. (Reuters) - From a wind-swept industrial site in the Sonoran Desert, Intel Corp., appears to be gearing up for battle.

Construction crews hammer away at an unprecedented $2 billion upgrade to one of Intel's two Arizona factories, preparing the world's largest chip maker to safeguard its lead in manufacturing from resurgent rivals and to put recent costly missteps behind it.

The stakes are high: If Intel can pull off its complex renovation of the 8-year-old Fab 12 plant, it could pioneer a much cheaper alternative to building chip fabrication facilities from scratch.

For Intel's top manufacturing executive, Robert Baker, the challenge is also to show that a technology powerhouse with a manufacturing staff of 45,000 -- six times the entire payroll of rival AMD -- can be a nimble innovator.

"Part of what I do is put the emphasis on how fast we respond," Baker, 49, said in a recent interview.

Intel needs to move faster than ever in its 36-year history after a series of product blunders in 2004, including a recall of defective desktop computer chips.

Meanwhile, Advanced Micro Devices Inc., Intel's smaller rival that made its name by making bargain-basement Intel knockoffs, has been gaining in consumer niches like chips for PC gaming and computer servers. And with help from partner IBM, AMD has honed its manufacturing, especially in factory automation.

"AMD is the perceived leader on the automation side, not Intel," said Risto Puhakka, vice president of industry forecaster VLSI Research.

But AMD is not the only threat. South Korea's Samsung Electronics Co. Ltd. has been spending billions on chip factory equipment and has impressed the engineering community with its manufacturing research. Texas Instruments Inc., which competes against Intel in cellular phone chips, is at work on a $3 billion advanced factory of its own.

From his office in a nondescript building overlooking alfalfa fields and dairy farms, Baker, Intel's senior vice president in charge of manufacturing, turns out strategies to keep Intel's factories in the lead.

Baker's group will lead the industry forward on three new technologies; the move to dual-core, or two-chips-in-one, microprocessors; the shrinking of chip feature sizes to 65-nanometers, small enough to be on the same scale as viruses; and the move to more productive 12-inch silicon wafers.

His challenge is to make the new ideas work in the real world. A team in Oregon develops the technologies, but Baker's group puts them to work in factories. Inside the engineering community, Intel's ability to churn out hundreds of millions of chips a year is half the basis of its reputation.

"Intel's biggest strength has been their manufacturing," said Charlie Glavin, an analyst with Needham & Co.

In an environment where an errant speck of copper or a faulty circuit layout can ruin millions of dollars worth of goods, the stakes are high. "If Intel's able to do this, they will get higher (profit) margins," Glavin said.

Part of Intel's gamble in Chandler is to try to convert an older plant to new technologies. Building new is easier -- but more expensive. The upgrade could save costs, but potential losses for a miscalculation are also high. If Baker gets it right in Arizona, factories around the world -- including Fab 22 next door -- could get similar treatment.

Fab 12 (fab is industry shorthand for fabrication facility) was one of the world's most advanced production facilities when it was built eight years ago. But the drumbeat of Moore's Law, the maxim named for Intel's founder that predicts the constant shrinking of transistors, the building blocks of microchips, had made it nearly out of date.

To upgrade the factory, the interior structure and layout of the entire facility must be changed, new tools installed, and the supporting infrastructure above and below the factory reworked. All new automation tools that shuttle pods of 12-inch wafers -- each worth up to $70,000 -- have to be installed.

The missteps in 2004 taught Baker a few lessons, mostly about the processes for decision making, internal communication, and planning -- all of which were changed in the interest of getting Intel "to not act like a big company, but act like a small company," Baker said.

But he is still committed to big changes, and he says that such a road always has rough spots.

"When you try to do state-of-the-art stuff, you're going to run into things," he said.

Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...