How Israel saved Intel

Five hundred employees and guests crowded under a white tent half the length of a football field at Intel's Santa Clara, Calif., headquarters as Chief Executive Paul Otellini put his company's newest line of computer chips through their paces.

"These are the best microprocessors we've ever designed, the best microprocessors we've ever built," Otellini said. "This is not just incremental change; it's a revolutionary leap."

Otellini's pronouncement relegated to obsolescence Intel's Pentium chip, which once powered more than 80 percent of the world's personal computers. That wasn't the only surprise last July.

A camera zoomed in on engineers in lab coats in Haifa, Israel. The video revealed that the chip Intel is counting on to recover from a battering by Advanced Micro Devices (AMD) wasn't invented in Silicon Valley. Instead, Intel is betting on a group of Israeli mavericks and a design bureau 7,400 miles away.

Shmuel Eden, former head of the Israel Development Center, where the new Core 2 Duo was created, says he's fed up with the perception that Intel's prowess is fading.

"People are angry," said Eden, who moved to Santa Clara in 2002 as sales manager for Intel's laptop chip division. "When I see something in the press saying Advanced Micro has taken our lead in technology, it hurts me personally."

Investors are hurting, too.

"I can't see Intel getting back to the market-share levels they used to have," said William Gorman, an analyst at Philadelphia-based PNC Wealth Management, which manages about $50 billion, including Intel shares. "They opened a window, and AMD took advantage."

Intel's share of the $33 billion computer-processor industry is the lowest in 11 years, according to Cave Creek, Ariz.-based Mercury Research. Profit plunged 42 percent to $5.04 billion last year as the company slashed prices after Intel's share of the personal-computer processor market slipped to 75 percent.

Advanced Micro, Intel's Sunnyvale, Calif.-based rival, unveiled its first server processor, called Opteron, in 2003. Since then, AMD has wrested customers away from Intel.

One, the Philadelphia Stock Exchange, switched its servers to Opteron in 2005 and then chose AMD-powered computers for its network of 600 desktop machines.

"When I find a technology that works for me, I'm going to stick by it," said Gene Peters, director of information services at the exchange. "I've no doubt that we made the right choice and will be with it for the next five or six years."

Sales, shares fall in '06

Without customers like Peters, Intel's sales dropped 9 percent last year to $35.4 billion, even as the PC market grew 10 percent.

Dell, Intel's largest customer, compounded the woes last May by ending its exclusive use of the company's chips.

Intel shares finished 2006 down 19 percent, the worst performance in the Dow Jones industrial average. That includes the 16 percent rise that began on July 27, the day Otellini unveiled the Israeli-designed processor. Intel shares have fallen 5 percent in 2007, as of March 27.

"I just can't get excited about what's going on," said Daniel Morgan, who helps manage $5.45 billion, including Intel shares, at Columbus, Ga.-based Synovus Investment Advisors. "Do they ever regain the prominence and bellwether status that they achieved during the '90s?"

That answer would be a definite no if not for the Israeli team, said Doug Freedman, an analyst for Greenwich, Conn.-based brokerage American Technology Research.

"They saved the company," Freedman said. "Without those new products, Intel would be in a lot more trouble."

Otellini's bet on the Israelis required a shift in thinking about how processors work — and how Intel markets them. Intel had always promoted the mantra that faster clock speed, the rate at which a chip executes instructions, was the key to measuring how well a computer performs.

Eden, who prefers a beret and a leather jacket to Intel's unofficial uniform of a blue shirt and khakis, says the Israeli team realized the danger in the industry's obsession with speed. For decades, making chips perform better had meant cramming transistors closer together and switching them on and off more quickly.

Hitting the power wall

After almost 40 years, the postage stamp-size devices contained millions of transistors flickering more than a billion times a second. The penalty was heat — as much as a small television set generates.

As the Israelis contemplated computer makers' designs for thin and light notebooks, they realized they'd need a fan thicker than the computer itself to keep the processors from cooking. That would make slim laptops impossible.

"We hit the power wall," Eden, 53, says. "We said, 'Guys, it cannot work anymore.' "

In Haifa, Rony Friedman was working on a chip for low-cost computers. It was slower than Intel's lineup — but it put out less heat. The trick was to persuade the rest of Intel that the company could sell a slower processor. Winning over Otellini, then head of the chip division, proved a turning point.

"We did it the Israeli way; we argued our case to death," Eden recalls. "You know what an exchange of opinions is in Israel? You come to the meeting with your opinion, and you leave with mine."

The Israelis' new chip, named Banias after a river in Galilee, got a break with Intel's push into notebook computers.

It hit the market in March 2003 as part of a package called Centrino. It let business travelers tote around PCs that had the same performance as their desktop models.

Banias anchored three years of 13 percent annual sales growth at Intel from 2003-05.

The trouble was that while Intel was working on laptop chips, AMD had persuaded IBM to use the Opteron processor for servers.

By the end of 2006, Advanced Micro had snatched about a quarter of the server market, up from about 4 percent before Opteron came out in 2003.

Intel's Type A personalities are rallying. In one 100-day period last year, Intel unveiled more than 40 processors, most based on the Israeli design.

Friedman now runs Intel design groups around the world.

"Now we're doing processors that should carry most of Intel's revenue," he said. "We can't screw up."