Technology

Is the Semiconductor Boom Too Much of a Good Thing for Intel?

It Fights to Meet a Flood of Demand -- And to Catch the Next Wave in Chips

Business Week

April 23, 1984

Despite an excellent longterm growth record, semiconductors have always been a feast-or-famine business. Economic downturns cause profits to disappear and capital spending to decline. During the most recent recession, however, Intel Corp. tried to prepare for the next boom by increasing its annual outlays for more production capacity. From 1981 through 1983, the industry's technology leader spent $275 million on chipmaking equipment and still managed to stay out of the red -- though just barely. But that is slim consolation to the herd of Intel customers who are seeing red because they cannot get enough products from the microprocessor pioneer.

Intel has been overwhelmed by exploding demand. Many frustrated customers charge that the company should have done a better job of managing growth. "Intel said they couldn't do miracles," recalls one irate customer. "So I threw the salesman out and we designed out their chip." The most extreme example of this deluge of orders is the microprocessor family that serves as the brain of the wildly successful Personal Computer from International Business Machines Corp. and its dozens of clones. Both the 8086 chip and its faster, lower-cost version, the 80186, are sold out for many months.

But management maintains that no company could have anticipated such a powerful surge in demand. The full extent of the problem suddenly became apparent late last fall, when a customer survey showed that 1984 demand for the 80186 would be six times what Intel had forecast in a long-range planning meeting just a few months earlier.

INCREDIBLE EXTREMES

Intel is "going to incredible extremes to meet demand," claims David L. House, general manager of Intel's microprocessor group. The first floor of offices at his group's Santa Clara (Calif.) headquarters has been converted into a factory, for example. Capacity for the 80186 is being expanded at an unprecedented rate -- "10 times that of any previous microprocessor," says Jack C. Carsten, senior vice-president. Although Intel is making twice as many of these chips as planned, says Chairman Gordon E. Moore, "that is only a third as much as the market wants."

Industry experts do not expect the company to catch up with demand before the end of the year, despite a doubling of last year's $107 million spending on chipmaking equipment. As annoying as this may be to customers, the effect on Intel's balance sheet is certainly salutary. If orders continue to pour in at current rates, says Moore, "we will have doubled in size by this time next year." Intel closed the books for 1983 on its first billion-dollar year, with a jump in net income to $116 million, nearly four times as much as the year before.

Because IBM's PC established the company's 16-bit microprocessors as an industry standard, "Intel has achieved its strongest competitive position since the mid-1970s and has oligopolistic positions in certain product lines," says John J. Lazlo Jr., who follows the semiconductor industry for Hambrecht & Quist Inc. In the exploding market for 16-bit microprocessors -- which Dataquest Inc. estimates will climb 55% this year, to $133 million -- Intel and the chipmakers licensed to produce its designs command a 70% share of all units shipped.

While the company's short-term future seems secure, some Intel-watchers are less sanguine about its long-range outlook. For example, the microprocessor leader does not yet have a strong candidate for the next-generation product -- 32-bit microprocessors that handle twice as much data at a time as today's 16-bit models. Intel trails both Motorola Inc. and National Semiconductor Corp., because its first 32-bit design, introduced in 1981, was a commercial flop. "Intel produced a dog," says Frederick L. Zieber, a Dataquest analyst, "but the next one could be a tiger."

Intel is also late in moving to what is rapidly becoming the most important integrated circuit technology -- CMOS, short for complementary metal oxide on silicon. This manufacturing process is about the only viable process for tomorrow's integrated circuits, which will cram hundreds of thousands, even millions, of transistors onto each fingernail-size chip. Without using low-power CMOS, which generates far less heat than other processes, such complex chips would not be practicable. "Motorola jumped on CMOS long ago, and then National came on strong. But Intel has been lagging," says market researcher Jack Beedle, president of In-Stat Inc.

BEST-KEPT SECRET

The company's diversification into systems built around its microprocessors has also proved disappointing. With much publicity, Intel entered this business on a broad scale in 1982, hoping it would insulate the balance sheet from the violent, cyclical swings in semiconductors. Systems account for about 25% of corporate revenues, but Moore acknowledges that "we are not completely happy with our performance in systems" -- either in market share or profit margins. Intel's share of the systems market is pegged at less than 4% by Alexander D. Stein, who tracks the computer industry for Dataquest. He calls Intel's systems "the industry's best-kept secret."

Critics say that Intel's systems business has never yielded the profit levels to which Intel has grown accustomed. "Some parts of Intel's semiconductor business have gross profit margins of more than 90%, whereas the systems business produces margins of 50% at the most," says Vaemond H. Crane, who headed Intel's systems activities until resigning last year. As a result, critics believe that the company's investments continue to be skewed toward chips.

Outsiders also wonder whether Intel's drive into systems will be sapped by IBM's investment in the chipmaker. One competitor argues that "IBM may not want Intel in parts of the systems business, and he who has the money has the clout." In 1982, IBM bought a 12% stake in Intel for $250 million in cash, subsequently increased its holdings to 19%, and has authority to buy 11% more.

IN THE PIPELINE

Intel executives are quick to respond to all these criticisms. Moore reports that a new 32-bit microprocessor "is in the pipeline for late this year." Unlike its ill-fated predecessor, it will be upwardly compatible with Intel's current 8086 family -- and will be produced in CMOS. Carsten points out that for almost three years, all new Intel chip designs have been developed exclusively in CMOS. The company may be shipping a negligible quantity of CMOS chips, Moore admits, but that is because "we've been busy making a family of [conventional] parts that the world is falling over itself to get. We have no need to be in CMOS for CMOS's sake."

Some customers are already beginning to notice Intel's new aggressiveness in CMOS. Stephen A. Albino, senior component engineer at Prime Computer Inc., says that "they seem to be making up for lost time." Even some chipmakers agree. "Intel has caught up with the technology leaders -- Toshiba and Hitachi -- in R&D terms, even though it is not yet shipping state-of-the-art product," says T. J. Rodgers, president of Cypress Semiconductor Corp.

But even if Intel is closing the gap in CMOS, an executive at a major competitor contends that Intel may have abrogated its position as technology leader. Intel, he explains, has decided to downgrade its emphasis on computer memory chips, which most producers use as the test bed for new production methods. Leslie L. Vadasz, an Intel senior vice-president, retorts that other chips now provide the kind of volume that memories provided, and Intel will use these products as its proving ground for new process techniques.

As for any impact on Intel's systems business by the IBM investment, Moore insists that it has had no effect. "We were adamant about that." As evidence of the arm's-length relationship between the two, Carsten points out that the portion of Intel's chip sales going to the computer maker has declined. It now stands at about 9%, down from 13.5% in 1982, although "IBM would probably like us to ship them twice as much."

Intel has fared better in some parts of the systems market than in others. One weak spot is the original-equipment-manufacturer (OEM) market in office automation. For example, Intel hoped to make a splash in 1982 with a low-cost data-base processor designed to speed data retrieval from large computers. At the time Intel President Andrew S. Grove termed the processor a "world-beater," but William W. Lattin, head of the systems group, now admits that it has proved only "modestly successful."

On the other hand, Lattin says Intel is enjoying "a great deal of success" in systems designed for resale by GEM customers in manufacturing and engineering markets. And last year, Intel launched two new general-purpose processors, costing $5,000 to $30,000, that the company believes will attract OEM buyers from both commercial and technical markets. Dataquest's Stein believes these new products "could go a long way toward making them a significant force, if they market them properly."

For now, though, the task of coping with excessive demand has top priority, and Intel is earning praise for the way it is dealing with the matter. Says Kenneth V. McKenzie, a senior analyst at Dataquest: "Intel is handling the shortage better than any other semiconductor company." Last summer, for example, the company instituted a "bookings control system" to help customers get an accurate estimate of how many chips Intel could supply, so they could, if necessary, line up chips from other suppliers.

More recent strategems aimed at easing chip shortages include:

* Subcontracting mature product lines. Intel is currently putting six chipmakers through qualification tests. Those that pass will produce circuits that Intel will package and sell as its own chips.

* Licensing customers with chipmaking facilities to make parts for their internal use. Intel is giving the necessary tooling to these customers, which include IBM and Commodore Business Machines Inc.

* Licensing additional components to long-standing "second-source" suppliers, or fellow chipmakers empowered to make and sell Intel-designed chips on the open market. These include Advanced Micro Devices and Harris in the U.S., Siemens and Philips in Europe, and Fujitsu and NEC in Japan.

Intel watchers generally applaud these initiatives, especially the subcontracting. Says former Intel manager Philip A. Kaufman, now president of Silicon Compilers Inc.: "It's a great strategy. They are giving the mature parts to other people so they can use their capacity to build the new high-margin ones."

To turn out more high-margin chips, Intel recently opened a new chipmaking plant near Albuquerque -- the first in the U.S. to build chips on 6-in.-diameter silicon wafers. Current plants use 5-in. wafers. The difference may not seem significant, but for a modest increase in wafer-making costs, Intel obtains a dramatic reduction in costs, because the number of chips per wafer practically doubles. "Intel is leading the way in 6-in. wafers. They are way ahead of everyone in that regard," says Gaynor N. Kelley, executive vice-president for operations at Perkin-Elmer Corp., which hopes to put some of its fabrication equipment in the new plant.

And a year from now, Intel's plant in Chandler, Ariz., will speed the flow of chips to customers by cranking up an automated line for putting the tiny slices of silicon into protective packages. Today, this is a labor-intensive job typically done in Mexico or Southeast Asia. But Intel plans to phase it back into the U.S. because, says John A. Ekiss, general manager of the Special Components Div., "the transportation content of packaging is now actually more than the labor content." Ekiss recognizes that the industry in general has not yet reached this conclusion, "but if we're right, if it gives us a significant cost advantage, then we'll be in a good position."  

GRAPHIC: Picture 1, CHAIRMAN MOORE: DOUBLED OUTPUT SUPPLIES "ONLY A THIRD" OF THE MARKET, LIANE ENGELIS; Picture 2, INTEL BOOSTS PRODUCTION BY FITTING TWICE AS MANY CHIPS ON 6-IN. WAFERS, BROWNE/PICTURE GROUP; Picture 3, GENERAL MANAGER HOUSE: "GOING TO INCREDIBLE EXTREMES" TO FILL ORDERS, BROWNE/PICTURE GROUP

Copyright 1984 McGraw-Hill, Inc.