Information Technology: Chips
Intel Inside Out
After a year of bloopers, can the chipmaker get its house in order?
Intel Corp. (INTC) has eaten a pile of crow this year. The $34 billion chip giant, known for its disciplined management and crisp operations, suffered a nightmarish string of product shortages, delays, and even cancellations that left customers and investors fuming. Case in point: After agreeing to use Intel's latest Pentium III Xeon processor in a new line of servers--and promising deliveries to customers--Compaq Computer Corp. (CPQ) was shocked when the chipmaker couldn't provide enough parts and buyers were left hanging. Turns out Intel had underestimated the growth of the PC market in 2000 and failed to lay on enough new plant capacity. "Intel is a good marketing machine, but they didn't put enough focus on production," gripes a Compaq official.
Ouch. Could this be the same Intel whose tight execution, relentless growth, and bountiful profits have made it an icon of techdom? For most of this year, it hasn't looked that way. First, pesky bugs forced it to delay delivery of crucial chipsets--key products needed to connect processors to the rest of a PC. Then it had to recall a million circuit boards because of a bad chip and eventually gave up on fixing the faulty component. Meanwhile, it suffered chronic product shortages through much of the year. And most embarrassing, in August it had to recall the latest version of its flagship Pentium III because the processor simply didn't work. It had been rushed to market to blunt rising competition from archrival Advanced Micro Devices Inc. (AMD) "Intel took its eye off the ball," says analyst Ashok Kumar of U.S. Bancorp Piper Jaffray. "Its execution this year has been anything but commendable."
The bloopers haven't gone unnoticed on Wall Street. Intel took hundreds of millions of dollars in write-downs and may have forfeited as much as $1.4 billion in lost revenues, according to analysts. That has led nervous investors, already shaken by tech-stock weakness, to slice a staggering $235 billion from Intel's market capitalization since August. It is now valued at $275 billion, some 45% off its peak. "Investor sentiment is extremely negative on the stock," Kumar says.GAWKY STAGE. The concern isn't just over the fiascoes of 2000. After all, despite the problems, Intel is still expected to turn in 16% revenue growth and a remarkable 49% profit increase for the year. What worries Wall Street is whether Intel can pull off its ambitious plans for 2001. That's when the company plans to switch to a new manufacturing technology, pump out millions of its new Pentium 4 processors introduced on Nov. 20, and roll out the complex, long-awaited Itanium superchip for big-iron computer servers. The pressure is on Chief Executive Officer Craig R. Barrett, whose reputation as a skilled operations exec has been tarnished during his two-plus years at the helm. The bottom line: "If they slip again, they're screwed," says analyst Jonathan J. Joseph of Salomon Smith Barney.
Getting a grip on operations isn't the only challenge. Faced with slowing growth in its core microprocessor business, Intel is scrambling to find new avenues for expansion. Barrett is pushing into areas such as networking chips and Web hosting, or running Web sites for other companies. These businesses now make up one-fifth of revenues and are growing at 50% per year. But the diversification is straining management and a culture that grew up designing and marketing one major product--the Pentium processor. "Intel is at the gawky adolescent stage," says analyst Nathan Brookwood of researcher Insight 64 in Saratoga, Calif. Even Chairman and former CEO Andrew S. Grove has confided to friends that he's worried about whether Intel can retain sharp enough focus on its cash-cow processors while tackling new and unfamiliar markets. Grove declined to comment.
Barrett is definitely feeling the pain. Intel's problems have spurred him to an uncharacteristic bout of public contrition. He says he is "embarrassed" by Intel's slipups and admits it took unwise shortcuts. "We dropped the ball," he says. That's especially galling for an executive who paved his 24-year path to Intel's top job by helping turn the company into a manufacturing machine with unmatched quality and 64% gross margins. Intel always has practiced "prudent and intelligent risk-taking," he says, but this year it tried to move too fast and without proper controls.
The steely CEO vows he won't let that happen anymore. Since the Pentium III recall, Barrett has stepped up an aggressive program to get Intel humming again. In biweekly executive staff meetings, he has berated his top lieutenants for sloppiness, insisting they adhere to Intel's rigorous planning methods. After all, he says, when things go wrong, "the enemy is usually us." Now he's tying management bonuses to more specific operational goals, such as hitting certain milestones and timetables. And he has gone to the stump, giving internal back-to-basics speeches and insisting on training throughout the company to reacquaint employees with techniques such as critical path analysis--a method for prioritizing work--that Intel has practiced successfully in the past. "We're getting back to what we know works," Barrett declares.SHUFFLING. That also means some midcourse corrections. Barrett has pumped up Intel's capital spending this year by 75%, to $6 billion, to ensure plentiful capacity going forward. To better focus the company's engineering efforts, the CEO in late September canceled a planned low-cost chip, code-named Timna, that had run into technical troubles and delays. And on Oct. 9, he shuffled top managers to oversee a sudden doubling in the production levels of the Pentium 4. The most prominent casualty: Intel's veteran head of processor design, Albert Y.C. Yu, who was given a new assignment to spearhead Intel's nascent efforts in optical chips. Insiders say the move was long overdue. "Albert hadn't added value in years," says a former Intel exec. Yu couldn't be reached for comment.
Nothing is sacred. Barrett even has begun publicly backing away from a crucial partnership. In 1996, concerned that conventional memory-chip technology was running out of gas, Intel hitched its wagon to startup Rambus Inc. (RMBS), which had developed an unproven alternative that dramatically boosts the connection speed between memory and the processor. Barrett now says Intel made a "bad bet" four years ago by committing to use only Rambus-type memory in all its new PC designs. The technology has proved pricier and more complex than anybody expected. So by next summer, Intel will deliver chipsets for the Pentium 4 that let it work with cheaper conventional memory, which has improved in the intervening years.
How did Intel get into such a mess? Most of its product delays and cancellations stem from difficulties implementing Rambus' technology--and to Intel's lack of a backup plan in case that didn't pan out. The capacity crunch traces to underinvestment in new plant capacity in 1999, when Intel slashed capital spending from a planned $5 billion to just $3.4 billion. And the rushed release of the buggy Pentium III--after an engineer neglected to run a crucial test on the part--was undoubtedly spurred by Intel's competitiveness and paranoia about AMD. The rival "is pushing Intel to do things it wouldn't normally do," chides chip-industry analyst Linley Gwennap of The Linley Group. "It's moving too fast and getting careless."NO TRACTION. That's not to say all of Intel's actions have been cavalier. In 1996, when Intel selected Rambus, analysts forecast that with Intel's backing, Rambus-type chips would command 50% of the memory market by now. That hasn't happened. Intel knew a year ago that Rambus wasn't getting traction, but it was so committed that it couldn't back down. To buy some wiggle room, it devised a "translation" chip that let its Rambus chipsets talk to conventional memory. Problem was, the translator didn't work, and Intel didn't figure that out until after it had shipped a million of them to computer makers. That forced a recall in May, followed by a $253 million write-down. Meanwhile, PC makers turned to Taiwanese rivals. The result: Intel's share of the $4.3 billion chipset business has fallen by up to 13 percentage points since 1999, estimates Mercury Research, costing the company an estimated $560 million in revenues this year alone.
Intel's capacity constraints cost even more. When planning its 1999 capital spending, Barrett says, Intel misread this year's PC market, predicting 10% unit growth instead of the 18% now expected. Sure, Intel was only one of many companies blindsided by strong demand in early 2000. "Apparently, we were all idiots at the same time," Barrett quips. But Intel also took some added risks. Under enormous financial pressure after a dismal 1998, when profits declined for the first time in a decade, Barrett and his team banked on a lower-cost approach to updating their chip plants. The scheme worked, and the savings were huge, but Intel lost vital capacity just as chip sales surged. "We didn't have enough buffer," admits Michael R. Splinter, executive vice-president for technology and manufacturing. "Not being able to meet demand was unbelievably painful."
Costly, too. Just as Intel ran into its chipset problems and capacity crunch, AMD hit paydirt with a sizzling Pentium III rival called the Athlon. The cheaper chip has helped AMD win business with nine of the top 10 PC makers and gain five points of market share this year, to 18%. Analysts say much of that gain, worth at least $800 million, might otherwise have landed on Intel's top line. AMD also managed this year for the first time to grab the chip-speed crown when it delivered the first PC processor to run at 1 gigahertz, or a billion clock ticks per second. Intel rushed out a riposte six months ahead of schedule and soon after shoved the faulty 1.13-GHz Pentium III out the door.RADICAL SHIFT. To regain momentum, Barrett is doubling the production ramp-up of the Pentium 4. And to bring off that ambitious goal, he's making a radical shift, assigning everybody involved with the product--designers, process engineers, and manufacturing personnel--to report in cross-functional teams to one executive, Splinter. Now, instead of running every decision up the flagpole and back down again, Splinter says, the Pentium 4 launch will be managed by more nimble groups. "We have to shorten the decision process," he says.
There's no doubt Barrett & Co. are fired up. Intel has always performed best when it's on a mission, and right now all 82,000 employees want to show the world that Intel hasn't lost its touch. During the worst of the crises, says Executive Vice-President Paul S. Otellini, "we were like a deer in the headlights." But at some point, he adds, people flip from being scared to being mad. Now, he says, "you'll see us bounce back like never before." If not, Barrett will have to ready himself for more painful apologies.By Andy Reinhardt in Santa Clara, Calif.Return to top