Science and Technology
Motorola's microprocessor dilemma
August 2, 1980
Motorola, which ranks third in world sales of integrated circuits, is in the midst of a critical debate on its strategy for the most important silicon chip of the 1980s: the new generation of microprocessors, which will control anything from robots to computer terminals.
Two main rivals, Intel and Texas Instruments (TI), have both decided on an expensive strategy for providing microprocessor systems based on these chips. Their systems approaches have as much in common with IBM philosophy as with the traditional practice of microchip manufacturers: to concentrate on making low-cost components and leave the customer to work out how to use them. Motorola does not know whether to follow Intel's and TI's suit.
The product under debate is called a 16-bit microprocessor. The first generation of microprocessors accepted instructions in 4-bit words: ie, in groups of four digital letters at a time. Next came 8-bit microprocessors, which could accept a longer set of instructions, made of eight letters at a time. Now comes the 16-bit (16-letter) microprocessor, which can accept as sophisticated an instruction as most users are likely to need.
The trouble is that many potential customers do not have the skills to design such a sophisticated device into their end-products (robots, computer terminals, etc). Some experts reckon the average cost of developing an application for a 16-bit microprocessor is $5m. This is a serious constraint on the marketability of the new microprocessor generation.
One problem is that the customer needs to design a computer board, containing not only the central microprocessor chip but also a whole gamut of supporting chips, in combinations that vary for each application. One solution is for the chip manufacturer to simplify the customer's task by assembling a range of off-the-shelf microprocessor systems, with combinations of chips that customers can simply plug into their products. Intel is particularly keen on this approach.
A second problem for the user is writing the enormously complicated software that is needed to give the microprocessor its sophisticated instructions. Intel and TI have been pioneering an expensive solution to this problem: commitment to a standardised range of software that can be shared by many applications, thus reducing the amount of software that has to be tailor-made for an individual application.
Motorola's dilemma is whether to follow these costly initiatives in order to assure its success in microprocessors. Its 16-bit microprocessor, called the M68000, is conceded by many to be the best designed product in this class, in terms of its basic architecture. That superiority is not yet reflected in sales figures, because the M68000 was only announced last summer, whereas TI launched the first 16-bit device three years earlier.
Motorola's hesitation about following companies like Intel and TI into the systems business reflects the history of its semiconductor division. The division's strength is in manufacturing the broadest possible range of semiconductor products, at the lowest possible cost, serving a wide range of customers. This strategy leads to high volumes and relatively low profit margins (15%, on average, before tax last year); Intel confines itself to a narrower market segment of innovative products that consistently earn profit margins over 20%.
Many top managers in Motorola's parent group argue that it should stay with its skills as a component supplier, remaining clear of the systems business. This view is reinforced by the experience of National Semiconductor, whose chairman, Mr Charles Sporck, told a recent conference that getting into systems (which now account for a profitable 20% of the company's turnover) involved ''a 10-year gestation period, and I wish to hell I hadn't started it''.
An idea of the expense of the systems approach is the fact that Intel is committing $200m for support of its 16-bit microprocessor. Texas Instruments reckons it has already spent more than this in support of its 4-bit and 16-bit microprocessor ranges.
It is only recently that Motorola has been able to afford the luxury of the systems dilemma. Its semiconductor division went into sharp decline in the late 1960s when a handful of key executives moved to Fairchild. An impressive recovery was engineered in the mid-1970s, aided by tough, TI-style management controls and a re-staffed product design team. The M68000 microprocessor is by no means the design team's only recent success. Motorola's reputation is changing from that of a ''me-too'' supplier of microchips to that of an innovator, a success that is obviously bringing its headaches.
If Motorola decides it cannot afford to follow Intel and TI in microprocessor support, it has contingency plans up its sleeve. Already, it is spreading the heavy costs of designing the many support chips needed for the M68000 through technology-swap agreements with Hitachi, Rockwell and Thomson-CSF: in exchange for details of the M68000 design, each of these partners is designing additions to the product family.
Another possibility being considered is to form partnerships with individual customers, to help design tailor-made systems for their specialised needs. Motorola would offer basic technology, and assist with circuit design and setting up a production line. On some occasions, it might act as a second source for a jointly designed product. Such customised systems would be more expensive than their mass-produced equivalents. But the end-products might be superior.
GRAPHIC: Graph, The 16-bit race, Dataquest, Cupertino and Frankfurt
Copyright 1980 The Economist Newspaper Ltd.