Doomsday deflected

Don't believe the Intel disaster scenarios-except one.

By Geoffrey James, photo courtesy of Intel Corp. -- Electronic Business, 3/1/2005

Bad news travels fast, especially when it's about one of the most important firms in high tech. Even a cursory Web search on Intel reveals a list of disasters as daunting as last year's series of Florida hurricanes. Public cancellation of high-visibility chip projects. Microsoft's working closely with closest competitor Advanced Micro Devices. A scathing internal e-mail from CEO Craig Barrett (he becomes chairman in May). Are arrogance and complacency taking their toll on the largest semiconductor company in the world? Is this the end of its dominance?

Yeah, right. Although Intel may have fumbled in the first half of last year, its financials remain strong, its market share stable and its productivity as high as ever. Intel continues to have close channel relationships with the PC vendors, and although its Microsoft partnership is uneasy, there's no sign that the Wintel era is drawing to a close. Seen in perspective, last year's turmoil at Intel, far from being a sign of gathering doom, is actually a process of self-correction. Rather than resting on its laurels, Intel is adapting to a world in which a chip's speed is less important than that chip's function.

For Intel the year 2004 started out bad and got worse. In February the company announced a delay in the Dothan family of mobile CPU chips, a long-anticipated upgrade to the Centrino product line. In May it confirmed cancellation of the Tejas and Jayhawk 64-bit processors, which were supposed to provide an upgrade path for the Pentium 4. In June it announced a recall of the Grantsdale Pentium 4 chip set, released about a month earlier.

And then came July. The company announced delays of both the 4-GHz Prescott Pentium 4 chip and the Alviso mobile chip set (a follow-on to the Dothan), along with a memory controller flaw in the Lindenhurst Xeon server chip. At the end of the month, Intel CEO Craig Barrett was so frustrated by the firm's fumbles that he sent a flame-o-gram to Intel employees, complaining, "This is not the Intel we all know, and that is not acceptable!" Barrett's e-mail was leaked to the press, producing a PR disaster.

The "Intel's in Big Trouble" interpretation of the above events goes something like this:
Intel, like many other high-tech firms, has gotten so big and inefficient that it can no longer get quality products out the door on time. Intel's channel partners, the PC vendors, will now jump to other CPU sources, especially now that Intel's longtime ally Microsoft seems to be favoring AMD. Because of this, Intel is now on the brink of losing market share, which is why the company needs a new CEO.

Needless to say, Intel executives don't buy the gloom and doom. "We had some slips in the first half of last year," admits Intel CFO Andy Bryant, "but since then everything has been going smoothly, and frankly we don't see big problems in the future." He points out that rather than experiencing a difficult year, Intel actually wrapped up 2004 with revenues of $34.2 billion-higher than the previous record, set in the year 2000.

"We've been moving to a model in which raw CPU performance is less important than the ability to provide the best performance with the lowest-possible power consumption."
-Anand Chandrasekher, vice president and general manager, Intel Mobile Platforms Group

More important, analysts who study Intel closely don't believe the doomsday scenario either. "It's only natural for the press to take potshots when Intel doesn't bat a thousand," says IDC semiconductor analyst Shane Rau. Dan Hutchinson, president of VLSI Research, is similarly skeptical. "These things go in waves," he says. "One writer picks up what he thinks is a hot story about Intel's failing, and the other reporters simply follow suit." This isn't to say that there haven't been big changes at Intel. However, far from heralding Intel's imminent demise, those changes are actually positioning Intel for future growth.

There are three facets to Intel's supposed doomsday scenario. The first, the notion that Intel is getting fat and lazy, simply doesn't hold water. Although conventional wisdom says that small, agile firms outperform dinosaur-like competitors, the situation is reversed in the case of Intel and AMD. In 2003 (the last full year for which SEC reports are available), Intel beat AMD in both revenue per employee and profit per employee, two common measures of productivity (see the graphic "Comparative Productivity of High-Tech Firms," below).

Why the spate of product delays, cancellations and recalls, then? Anand Chandrasekher, vice president and general manager of Intel's Mobile Platforms Group, credits last year's problems to a change in the way the company views its value proposition. "We've been moving to a model in which raw CPU performance is less important than the ability to provide the best performance with the lowest-possible power consumption," he says. That transformation required the cancellation of some projects that had once been considered strategic.

Chandrasekher isn't blowing hot air, according to IDC vice president Roger Kay. "Intel has been running smack against the thermal barrier," he says. "Getting ever faster speeds and feeds was simply making its chips run too hot." Intel's solution is to shift gears away from the company's current architecture, which uses a single CPU on a chip, and adopt a dual-core architecture with two CPUs on each chip. The dual-core architecture is intended to provide more overall processing power without generating as much heat as a single-core chip with comparable functions. (For more on dual-core architectures, see "Packing a One-Two Punch October 2004.)

Intel's embrace of dual-core doesn't mean that the company was failing but only that, like everyone else, it was reaching the limit of single-core technology. "People joke about cooking a chicken inside a Pentium 4 PC," says Hutchinson, "but the Apple Macintosh G5 runs half again as hot and has to be water-cooled like an automobile." Rather than opting for increasingly Rube Goldberg heat dissipation methods, Intel wants to use the real estate more efficiently. "Intel already has the lowest ratio of heat to transistors in the industry," says Hutchinson, "but it needed a new approach to overcome the limitations of its current architecture."

One could argue that Intel would have been wiser to have changed directions without a flurry of cancellations and subsequent bad publicity, but public gaffes are inevitable at Intel, because the firm is extraordinarily candid about its future plans, according to Hutchinson. "People talk about the death of the Tejas as if it had been some big disaster, but that chip never even taped out," he says. "If Intel hadn't been telling the world about it for years, nobody would have even known the project was gone."

The second element of the doomsday scenario is the notion that Intel's delivery problems will cause PC vendors to further embrace alternative CPU sources, such as AMD. However, although AMD's low-priced CPUs have firmly established themselves at the low end of the PC market, Intel's not exactly hurting when it comes to market share. Although lower-priced x86 clones have been available for more than a decade, Intel still commands four fifths of the PC market. In fact, Intel's market share grew almost an entire percentage point (from 80.6 percent to 81.7 percent) in the first nine months of 2004, according to IDC (see the graphic "Worldwide PC Processor Unit Market Share by Vendor," below). More important, that market share gain was at the expense of AMD, which declined from 16.5 percent to 15.5 percent.

"When I ask Dell executives privately what they think about AMD, they shrug and tell me that they prefer doing business with Intel."
-Roger Kay, vice president, IDC

Intel's 80 percent market share is sometimes compared, unfavorably, to Microsoft's nearly total dominance of PC software. However, Microsoft has never had to cope with a competitor that offered a functionally identical product, whereas Intel has had to deal with the likes of AMD, Cyrix, and Transmeta. Rather than comparing Intel's market share with Microsoft's, it's more revealing to compare the market share of Intel's oft cloned CPUs with IBM's oft cloned PCs. Twenty years ago, Intel and IBM both owned about an 80 percent market share of their respective hardware segments. Today Intel is still at 80 percent, whereas IBM . not so much.

Intel's market share resilience is the direct result of the company's investments in marketing and branding, according to Rod Keller, who, until late 2003, was the head of Toshiba's U.S. PC business. Although Intel's chips are generally more expensive than AMD's, it can be more economical to use Intel's, because Intel rebates some of the extra expense in the form of joint marketing dollars. "A PC company might save a few dollars buying from AMD," says Keller, "but Intel will spend tens of millions of dollars to drive demand for your PCs, offsetting any potential savings you might get from a lower-cost CPU.

Intel's joint-marketing investment isn't chump change. The company regularly reimburses PC vendors between 50 and 70 percent of the cost of advertising, provided the ads feature the Intel logo. "Intel was extraordinarily generous and helpful when we were launching the ToughBook brand of ruggedized PCs in the U.S.," says John Harris, former marketing vice president at Panasonic's U.S. PC division. "It even implemented an automated system so we'd get reimbursed within a few days of running the ad."

Intel's joint-marketing programs have not only kept vendors in the fold but have also helped build Intel's brand. In its first 10 years, the "Intel Inside" campaign catapulted the company's name from a techie buzzword into one of the world's 10 most valuable brands, according to Interbrand, a company that focuses on brand marketing. Another branding company, Intangible Business Limited, says the Intel brand, even during the PC buying slump of 2001, was worth an estimated $35 billion. That's billion with a b.

Intel's brand awareness translates into customer brand preference, allowing Intel to charge more per chip than its competitors, which can sell a PC with an Intel Inside logo at a slightly higher price. That uplift keeps Intel's margins buoyant and allows Intel to "rebate" extra margin to the channel. The channel's advertising drives the sale of more PCs (and more Intel chips) while making Intel's brand more visible to the public. It's a strategy that Intel intends to continue for the foreseeable future, according to Bryant. "We are investing heavily in coadvertising, to make sure PC ads communicate the right message," he says.

Intel also spends money directly to build its brand and its subbrands, such as the Pentium and the Centrino. In 2004, for example, Intel spent $4.7 billion on sales, marketing and administration, more than AMD's gross revenue in the year 2000, the last year AMD made a profit. (As of this writing, AMD has not announced its 2004 earnings.) The advertising campaign for the Centrino chip set alone, for example, was budgeted at $300 million in 2003, according to Intel spokesperson Chuck Mulloy. And it's been money well spent, according to Rau, because the Intel brand, by itself, can sometimes move products off the shelf. "I recently received a Costco catalog with a coupon for a 'Centrino Notebook,'" he says. "You practically had to use a microscope to see the PC manufacturer's logo in the ad."

"Microsoft seems to be more afraid of irritating Intel than the other way around. The last thing it wants to do is to push Intel further into the Linux camp."
-Rob Enderle, principal analyst, The Enderle Group

Although most PC vendors have cut deals with AMD, Intel still has an exclusive arrangement with Dell, the industry bellwether. Some doomsayers, however, make much of Dell's recent public noises about considering AMD. But that's nothing unusual, according to IDC's Kay. "This surfaces every year," he says, "but when I ask Dell executives privately what they think about AMD, they shrug and tell me they prefer doing business with Intel."

The third thread in the doomsday theory is that a split between Intel and Microsoft is imminent. Under this scenario, Microsoft gets chummy with AMD and starts adding features into its operating systems that support only AMD chips. Microsoft's recent refusal to fully support Intel's Itanium and its support for the NX bit, a security feature in AMD's chips that is lacking in Intel's product line, lend credibility to this scenario.

However, viewing Microsoft's relationship with AMD as a major threat to Intel discounts the fact that Microsoft and Intel, like a Hollywood couple, have always had a love/hate relationship, filled with highly public drama and conflict. "They're constantly in competition to see who's the biggest gorilla on the block," says veteran Microsoft-watcher Rob Enderle, principal analyst for the Enderle Group.

For example, 10 years ago, Intel and Microsoft were supposedly at each other's throats over Microsoft's support of RISC in Windows NT. More recently the firms have squabbled over Intel's financial backing of Linux, as shown by, for example, Intel's investment in the Linux-based clustering software company Scali and funding of the effort to defeat SCO's anti-Linux lawsuit.

Despite the public squabbling, though, IDC's Kay believes that a real rift between the two firms is unlikely. "Too much of their revenues is tied up in each other's product offerings," he says. "They may flirt with competitors, but this is a marriage, for better or worse." IDC's Rau sees Microsoft's courting of AMD and Intel's fondness for Linux as merely an example of the two firms' hedging their bets. "Intel still faces competition from other CPU architectures," says Rau, "It's only smart to make sure that its processor architecture can be the basis for an alternative platform."

Intel and Microsoft must also work together to secure developing markets for the PC. As PCs penetrate the third world, the threat to Intel isn't so much AMD as Taiwan-based Via Technologies, according to Enderle. "Via is the mini-Intel, and its products are much more cost-effective," he says. Citing the example of PC companies that have failed to establish themselves in China, Enderle observes, "experience shows that big U.S. technology brands can be defeated when the local product is significantly cheaper."

But Via also represents a threat to Microsoft, according to Dave Mack, president of analyst firm Technology Business Research. "The third world is looking for a PC price point of around $200, and China has shown a marked preference for Linux," he says. "If Microsoft and Intel don't work together, they could both find themselves locked out." In short, Microsoft needs Intel just as much as Intel needs Microsoft. "Microsoft seems to be more afraid of irritating Intel than the other way around," says Enderle. "The last thing Microsoft wants to do is to push Intel further into the Linux camp."

Where Intel and Microsoft might actually part company is in their approaches to non-PC markets. Microsoft, for example, recently abandoned Intel CPUs for future versions of its Xbox game console. However, this decision probably reflects not pique with Intel's Linux strategy but, rather, Microsoft's desire to eventually leverage IBM's new Cell chip, which will be optimized for interactive computer games. Cary Snyder, business and technology analyst at Semiview.com, sees Microsoft's move as shrewd but far from relationship-breaking. "IBM is well on its way to dominating the emerging high-end consumer processing market with its Cell-based processing architecture," he says.

Boris Petrov, managing partner of the Petrov Group, believes that the Cell could potentially pose a threat to the Wintel duopoly. "Although in the past, threats to Intel haven't panned out, this time I think the company should be concerned," he says. However, IDC's Rau remains skeptical. "It's not yet clear how IBM is going to position the Cell," he says. "There are still a lot of unknowns." IDC's Kay believes that almost impossible hurdles would have to be overcome for the Cell to become a major player in the PC market. "It would be nearly impossible to dislodge the incumbents at this point," he says.

Intel executives, for their part, don't seem particularly worried about the Cell. "We heard this same talk a few years ago with Transmeta," says Bryant. "Low-power CPUs were supposed to turn Intel into toast, but we were able to launch viable products to capture that space." Far from worrying about disruptive technology from Intel's competition, Bryant insists that his main concern is the world economy. "Much of our growth depends on expanding into areas of the world where PCs aren't yet common," he says. "Any major upset in those regions might affect our worldwide shipments."

To the doomsayers, though, Bryant's nonchalance rings hollow. After all, didn't Barrett castigate Intel employees? Surely that's the definite sign that a company is in trouble. And what about the subsequent announcement of his successor, Paul Otellini? Isn't that evidence of dissatisfaction with his performance?

On the contrary. Rather than a sign of weakness, Barrett's memo is a clear manifestation of a strong company's taking strong medicine to correct its problems before they get out of hand. Think about it. Beleaguered CEOs aren't known for airing dirty laundry. Instead, they trot out weasel words, half-truths, and rumor denials. Other than its becoming public, the Barrett memo is pretty much business as usual at Intel, according to Chandrasekher. "We have a postmortem culture in which we brutally dissect everything that happens, both good and bad," he says. "Craig was simply articulating internally that we need to do better in the future."

IDC's Rau similarly sees the Barrett memo as the opposite of a management swan song. "Intel's corporate culture is aggressive and blunt," he says. "The language in the e-mail, if anything, is tepid compared with the remarks one might hear at a typical Intel meeting." As for Barrett's supposed "disgrace," boards of directors do not generally punish CEOs who bring in record revenues. And if they did, they'd probably find a better way to do it than making Barrett the new chairman of the company.

Intel's mid-January reorganization might justify a final gasp of pessimism, because troubled companies often restructure. In this case, however, the restructuring simply reflects Intel's dual-core strategy. The virtual elimination of Intel's desktop group suggests that power-efficient "mobile" chips, rather than power-hungry "desktop" chips will become Intel's lead products, according to Enderle. "The speculation is that in the future, desktop computers will start getting a modified laptop processor to cut noise and heat," he says. In any case, it's not unusual for a new CEO to start a tenure with a bout of musical chairs. "I don't think it's a sign of trouble," says Rau. "I think it's [new CEO] Paul Otellini's reorganizing for a new way of doing things."

In fact, the only really notable feature of the restructuring is that there's no identified COO, which means that Intel has not decided who will be CEO in the future, according to Adam Parker, a senior research analyst covering the semiconductor industry at the investment firm Sanford C. Bernstein & Co. However, many companies function quite well without a COO, and it may be a little soon for the company to identify an heir apparent.

For their part, Intel executives aren't troubled by the spate of bad press. "Big, successful companies are natural targets," says Bryant, "so I can't really blame the reporters who've been saying bad things about us." For Bryant, the best course is to take it all philosophically. "I guess it's because we're in America and people here simply can't resist rooting for the underdog."

What's the prognosis for Intel-is its current stumbles just temporary or a sign of permanent shift in the industry? Share your thoughts at feedback@eb.reedbusiness.com.

Geoffrey James has been messing around with Intel chips since the 8080 and is also a frequent contributor to Electronic Business.