A chip waiting to explode
Customers were confounded by the complexity of the 16-bit microprocessor chip
July 26, 1982
Three years ago, Intel Corp. loudly proclaimed that it had sewn up the market for the latest -- and what seemed destined to be the hottest -- generation of the so-called computer-on-a-chip: the powerful 16-bit microprocessor. The microprocessor pioneer was convinced that its commanding technological lead would translate into a dominant market position from which it could not be dislodged, and many rival chip makers tended to agree. But customers confounded both Intel and its rivals. Electronic-equipment makers found that designing the complex chip into new products was far more difficult than anyone had expected.
As a result, the Santa Clara (Calif.) chip maker slammed into a host of problems that have stunted the growth of the 16-bit microprocessor market and changed the competitive picture dramatically. Among other things, competitors gained the breathing space they needed to stay alive in this market. Intel's two closest rivals, Motorola Inc. and Zilog Inc., have jumped in with 16-bit micro processors that outperform Intel's original design. Even National Semiconductor Corp., which has been having a tough time getting its microprocessor effort moving, has just unveiled an innovative 16-bit entry.
Now that the 16-bit microprocessor market is taking off -- market researcher Dataquest Inc. estimates that worldwide sales will climb almost 70% this year to $110 million -- Intel finds itself with a real battle on its hands. Dataquest estimates that Intel accounted for 75% of 1981 sales of so-called second-generation chips -- the improved designs that it launched in 1978 with its 8086 chip -- vs. 54% of all 16-bit devices. But by 1990, forecaster: see Intel splitting 90% of an estimated $1 billion 16-bit business with Motorola. The remainder would be divided among Zilog, National Semiconductor, and Texas Instruments Inc. That projection includes only microprocessors and their peripheral chips, not the lower-performing microcomputer chips that combine those circuits onto a single chip.
These realities are triggering major changes in strategies among chip makers who are chasing the 16-bit market. By now, demand for the new microprocessors was supposed to be exploding. Because these newest microprocessors operate far faster than older designs and can address far more memory, Intel and the other microprocessor makers assumed that there would be no contest once their customers were ready to choose the electronic "brains" for such new products as word processors, communications gear, robots, and small computers. To them, counting on older 8-bit microprocessors to match the performance of 16-bit chips would be like expecting a car with a four-cylinder engine to match the power of one with a V8.
Despite these advantages, worldwide sales last year of 16-bit microprocessors and their associated support chips were only fractionally ahead of 1980 -- up from $63 million to $65 million, a long way from earlier forecasts of a $250 million market in 1981. At Intel, the alarm began clanging about 18 months ago. Dave House, general manager of Intel's microprocessor and development systems divisions, recalls that "we had to go to a crisis mode to find out what our customers' problems were."
Intel found that it had grossly underestimated the extent of help that its customers needed. The company had anticipated that developing 16-bit applications would cost users unprecedented amounts of money and effort, so Intel had planned to spend an eye-popping $200 million to develop application aids. But it found that these efforts were not nearly enough. "Customers needed higher-level tools than we originally imagined," admits House. "This is the fourth year since we introduced the 8086, and our new development tools are just now coming along."
"We all overestimated the ability of the market to absorb 16-bit technology," acknowledges Amaury Piedra, vice-president for international business at Zilog, the first company to unveil a second-generation design to compete with Intel. In spite of the technical excellence of its Z8000 chip, however, Zilog today has less than 7% of sales of second-generation units. "Until last year, we did not provide as much software support and development tools as we should have," concedes Bernard V. Vonderschmitt, general manager for components at Zilog. "Engineers who have designed equipment around 8-bit microprocessor systems are generally comfortable with them and are reluctant to use the newer 16-bit devices."
The biggest hangup, says Intel's House, is in programming -- writing the instructions that enable a chip to control a word processor or video game, for example. Instead of the 12 to 18 months typically needed to design the simpler 8-bit units into a product, House explains, "it's been more like 1 1/2 to 2 1/2 years for 16-bit applications." Intel responded by contracting with software specialists such as Microsoft and Digital Research Inc. to develop programs for its 8086 chip.
A resilient market
Yet even as Intel was striving to help customers cope with the complexities of that chip -- it packs 29,000 transistors, more than three times the number of its 8-bit predecessor -- the competitive pressures forced the company early this year to announce a faster, still more complex version, the 286. More recently, Intel announced another 16-bit design, the 186, that reduces the number of support chips needed to make it work, since much of the circuitry that usually goes onto peripheral chips has been squeezed onto the microprocessor chip.
Since a typical microprocessor application requires four or five peripheral chips, each costing more than the central processor, Intel expects the economic benefits of the comparably priced 186 to persuade more users of 8-bit chips to trade up to 16-bit models. The company wants to tap the exploding 8-bit market because it shows no signs of dying out in this decade. Dataquest estimates that 8-bit microprocessor sales last year increased 60% to $590 million and could climb 45% this year to $860 million and by 1987 hit $2.7 billion.
For Zilog, the largest supplier of 8-bit chips, the unforeseen resilience of this market is welcome, and the company is putting renewed resources behind the chips. Late next year, it plans to launch a high-performance, 8-bit microprocessor called the Z800. Industry experts suggest that next year could be too late, but Vonderschmitt argues that "an upgraded 8-bit product will benefit from all the software and all the peripheral circuits that already exist."
Because peripheral chips are so vital, Motorola knows it must catch up with Intel's huge lead to lure users to its 16-bit entry, the 68000. To hasten this effort, the big Phoenix chip maker has parceled out development of several peripheral circuits to other semiconductor companies licensed to produce the 68000. This expensive push will take 12 months more to complete, but Kenneth V. McKenzie, senior research analyst at Dataquest, believes that "on paper, Motorola and Intel are now evenly matched in peripherals."
Unlike the other competitors, Motorola claims not to be surprised by the sluggishness of the 16-bit market. "We never expected the 16-bit microprocessor business to take off exponentially," says William G. Howard Jr., vice-president for technology and planning at Motorola's Semiconductor Group. So even though the company is behind its original marketing plan because of the recession, he asserts that the "pause is not a bit disconcerting." Still, the recession and the tardy takeoff of the 16-bit market "are the best things that could possibly have happened from Motorola's standpoint," adds Dataquest's McKenzie. "Customers that have already chosen Intel's or Zilog's chip have had time to reexamine the alternatives."
Of all the strategic adjustments being made by U.S. chip makers, the one taking form at Texas Instruments is the most radical. Although a leading supplier of the early 16-bit parts, TI is an insignificant factor in the second-generation market. Instead of tackling Intel and Motorola head-on, the Dallas company has decided to bypass customer resistance and go one-up on Intel's 186 chip. Rather than selling a microprocessor witj just part of the necessary peripheral circuitry squeezed onto the logic chip, the Dallas company will emphasize single-chip microcomputers -- designs that include all of the supporting circuits and software needed to "take on the functions of intelligent subsystems specialized in, say, speech processing or file management," explains Walden C. Rhines, advanced development manager of TI's semiconductor group.
National Semiconductor, which finally introduced its 16000 chip in May, hopes to make up for its lateness with what F. Joseph van Poppelen, vice-president for marketing, calls "a better mousetrap." The company spent $38 million over five years to develop the chip, which features an advanced memory-management technique that until now has been used only in computers, never in microprocessors. While the advanced chip is certain to be welcome in market niches that require ultrafast performance, most experts concur with former Zilog President Federico Faggin's verdict: "Far from too little, but definitely too late."
Even Japanese chip makers, which have never been serious contenders in advanced microprocessors, may now have a chance to penetrate the 16-bit market as its life cycle stretches out far longer than had been expected. Notes Zilog's Faggin: "The Japanese [now] have more time to catch up -- and a greater incentive to do so."
The danger of this, some observers believe, is that if the Japanese put the same pricing pressures on the 16-bit market that have wreaked havoc in memory chips, profits could be cut to the point of jeopardizing U.S. dominance. The Japanese can afford to squeeze prices, explains Dataquest's McKenzie, "because they have licensed or reengineered U.S. microprocessor designs. They neither pay for R&D nor bear the initial marketing costs."
At the moment, Japan's share of the microprocessor business is small but growing rapidly in selected areas. For example, Japanese suppliers have focused on 4-bit microcomputers, a 10-year-old design, because these products are widely used in consumer electronics. Last year, for the first time, Japanese companies led worldwide sales of 4-bit chips, shipping 56% of total units.
The best way for U.S. microchip suppliers to maintain their lead, says Faggin, is "to step up the rate of innovation." To do that, chip makers now recognize, means that producers must provide better tools to help their customers keep pace with semiconductor technology. Intel President Andrew S. Grove puts it this way: "We have got to remove the technological cork for our customers so that the microprocessor genie can get out and expand."
GRAPHIC: Charts 1 and 2, Intel may lose its grip on 16-bit microprocessors, Data: Dataquest Inc. and BW estimates, Charles Rehwinkel
Copyright 1982 McGraw-Hill, Inc.