Intel pins its future on Merced

PALO ALTO, Calif. -- Intel has been at the crossroads before.

Two decades ago, with great fanfare, the company introduced a new microprocessor heralded as the chip that would reshape computing's future. It did -- but not in the way that Intel intended.

After six years in development, the i432 microprocessor became one of the great disaster stories of modern computing. A misguided effort to bridge the gap between hardware and software, the i432 ran five to 10 times more slowly than its competitor, the Motorola 68000, and was quickly withdrawn from the market. Serendipity rescued Intel from that debacle. When executives realized that the project was in trouble, they rushed to staff a stopgap effort that became known as the 8086 chip. Two engineers took just three weeks to design it, and the 8086 was rushed into manufacturing in 1978, just a year after its conception. From those humble and hurried roots came the computer chips that today power 85 percent of the world's PCs -- not to mention one of the world's most desired and hard-charging stocks. Now, to maintain its dominance of the chip business, Intel is again preparing to make a huge technological leap -- one that turns its back on the architecture of the 8086 and its progeny, up through today's Pentium processors. And given the volatility of the technology business, it is not an exaggeration to say that the future of Intel is very much at stake in how it manages the transition. Next year, the chip maker will begin shipping its Merced microprocessor. Code-named for the river that flows through the Yosemite Valley, Merced is the first of a new generation of chips that will permit the company, the world's largest maker of semiconductors, to directly attack a market that so far has eluded it: the very high end of the computer industry, made up of the mainframes that run the operations of large corporations, the exploding world of Internet Web servers and the supercomputers and workstations used by scientists and engineers. Intel is betting on a future in which the market for microprocessors is increasingly fragmented -- and in which the company has a distinct offering for each fragment: the superfast Merced at the high end, a growing spectrum of styles of Pentiums in the PC market and the ARM microprocessor, recently acquired from the Digital Equipment, in the low-end market for consumer appliances. The notion is that high profit margins on Merced chips can offset increasingly narrow margins in the competitive middle and lower reaches of the chip business. "Intel's view is that as the market is getting more segmented, we're adapting our products to serve different applications," said Steve Smith, the Intel executive in charge of the Merced business. But as the i432 disaster underscored, things do not always go as planned, even for the world's most admired companies. And there is a more troubling possibility, as well: By confining the Merced technology to the top end of the computer market, Intel may be vulnerable to competitors that will offer the same technology -- at far lower costs -- in other markets. If the next big wave in digital technology puts powerful computer brains in products like TV set-top boxes, smart telephones and hand- held computers -- where Intel is weakest -- the company's strategy with Merced could leave it beached. Already there is evidence that Intel has largely been shut out of the market for set-top boxes, for example, and its chips have proved to be too expensive and power-hungry for cellular phones or hand- held computers. The ARM chip may offer the company a way to be competitive -- but not with the monopoly pricing advantage it has enjoyed in the PC industry. The worst-case analogy would be to IBM. In the mid-1980s, the computing titan seemed unassailable. But by attempting to control the PC industry in a way that protected its profitable mainframe business -- in other words, by focusing on the high end of computing rather than the mainstream -- IBM in a few short years became a follower instead of the industry leader. Few people are expecting Intel to meet a similar fate. "The Merced is going to have a huge impact on the computer industry," said Larry Smarr, director of the National Center for Supercomputing Applications in Champaign, Ill. "The enormous volumes that Intel can generate will drive the current RISC microprocessors off the desktops over the next five years." No one has a better understanding of the pace of change in computing, and the risks inherent in each new generation of technology, than Andrew Grove, Intel's chairman and driving force. "When a change in how some element of one's business is conducted becomes an order of magnitude larger than what that business is accustomed to, then all bets are off," he wrote in his 1996 best seller, Only the Paranoid Survive (Currency/Doubleday, $27.50). "There's wind and then there's a typhoon, there are waves and then there's a tsunami." Merced has the potential to create that kind of computer industry tsunami, and Grove obviously believes he knows how to ride it: cautiously. Asked recently how long it would take for the chip to become the heart of the company's product line, he said: "It will be very slow. Not in the next five years. I don't see Merced appearing on a mainstream desktop inside of a decade." There have been persistent industry rumors of a struggle within Intel over how aggressively the company should push the Merced technology into high-volume products. And Intel made similar statements in 1993, when it introduced the Pentium microprocessor. That chip, too, was supposed to be just for servers and high-end workstations, but the desktop market embraced it quickly. Today, Pentium chips have found their way into PCs that cost less than $1,000, and Intel and its customers market them like cars or cigarettes. Whatever the truth of reports of dissent -- Smith acknowledged that Intel had "fought through" how to position Merced -- the chip is almost certain to quickly reshape high-end computing, a more than $60 billion market that until now has been dominated by International Business Machines, Sun Microsystems, Silicon Graphics, Digital and Hewlett-Packard. Indeed, that reshaping began even before Merced was named, when Intel entered into an alliance with Hewlett-Packard, a Silicon Valley neighbor with deep experience in high-performance computing, to develop a next-generation chip. The Intel-Hewlett alliance already appears so daunting that a year before Merced's scheduled debut, many of Intel's other competitors appear to be wavering. Not only has the chip giant co-opted one of its fiercest competitors by allying with Hewlett-Packard, but Silicon Graphics has said that it plans to build computers based on Merced, raising questions about the future of its MIPS microprocessors in the supercomputer and workstation markets. Prospects for Digital's Alpha processor, meanwhile, are in doubt because the computer maker has been acquired by Compaq Computer, Intel's best customer. Even Sun, famous in Silicon Valley for its iconoclasm, has indicated that it will develop a version of its Solaris operating system for Merced, leaving some analysts wondering whether the workstation maker can continue to press forward with ultrafast versions of its SPARC chips. Given all that, analysts say that only IBM, with a business of $20 billion to $25 billion a year based on the Power PC family of processors, has the scale and semiconductor manufacturing capabilities to stand up to Intel. But even IBM, with the world's second-largest personal computer business, will remain one of Intel's best customers. Few competitors are willing even to criticize Merced aloud, fearing Intel's pervasive influence in the computer industry. "Sure, I think there are shortcomings in their design, but I'm not going to say so publicly," one Silicon Valley computer designer said. "That would be like walking around in Iraq before the Gulf war carrying an American flag with a bull's-eye on my back."

Copyright 1998
Provided by ProQuest Information and Learning Company. All rights Reserved.