Technology: The cold facts about hot chips: Intel's problem as a world-beating profit machineis how to keep the returns rolling in

THIRD QUARTER results published last week by Intel, the giant US semiconductor manufacturer, will force analysts to revise upwards their full-year expectations of what is reputedly the world's most profitable large company.

Last year's total sales were dollars 5bn ( pounds 3.3bn); sales in the first nine months of this year were dollars 6.39bn, comfortably ahead of the full-year figure for 1992, making the analysts' forecasts of dollars 8bn for the full year look conservative.

Intel's fourth quarter is traditionally one of its strongest and the company looks set for another record year. But how long can such growth continue?

The statement to shareholders by Andy Grove, chairman and chief executive, shows what thin ice the company is skating on. While only 'hundreds of thousands' of the company's new Pentium microchips will have been shipped this year, 'millions' would be sold next year, as Pentium increasingly replaces the company's 486 chip in the personal computer market, he said. In fact, the future of the Pentium chip worries Mr Grove deeply.

As Intel has grown (it invented the world's first microprocessor in 1971 as a three- year-old start-up company), it has become heavily reliant on particular technologies and products. Its success in the personal computer market (for years the company was the sole supplier of the chips that are the 'brains' in IBM-compatible PCs, which means most PCs) polarised its prospects even more.

With margins on its four- year-old 486 chips shrinking as competitors'clone' their own versions of it - 486 chip prices have just come down from dollars 395 to dollars 311 - the company needs to move computer buyers' aspirations upmarket, to where only Intel produces a chip.

It is a familiar problem for Mr Grove, who has been with the companysince its foundation in 1968 by electronics engineers who broke away from other semiconductor companies. Every successive chip generation has posed the same problem, but this time the stakes are higher.

Even so, it might be thought surprising that Mr Grove considers the transformationof Pentium into the workhorse of the PC market as 'the biggest medium-term challenge' that he faces at Intel. The reasons soon emerge.

The company does not disclose research and development costs for individual chips, but it is certain that Pentium will have broken all R&D records. Chip development is expensive: the 1993 R&D bill alone will exceed dollars 900m, according to predictions earlier this month by staff at the company's headquarters in Santa Clara, California.

This is split between various projects, including the two successors to Pentium, P6 and P7, now on the drawing-board. But it is the capital-equipment expenditure on Pentium that is breathtaking. Building and equipping factories to make the chip has cost dollars 5bn to date, according to Mr Grove. Yet Intel's own forecasts concede that buyers will remain hooked on the much cheaper 486 chip for many years. Sales of the486 are still growing strongly, and may not peak until 1995 or even later. Pentium chips cost twice as much, and may well deliver less than twice the performance - if they provide any improvement at all - in some PC designs.

Intel microchip engineers are bitterly attacking PC manufacturers, claiming that they are deliberately switching off many of the Pentium's new features in order to avoid redesigning their systems. Unless Intel can win these arguments quickly, it will lose market share to competitors developing their own Pentium-killer.

Mr Grove's strategy is to reduce the company's dependence on successive generations of PC chips, hopefully by identifying the next 'killer' product.

'Our culture is still that of a start-up,' he claims. He is determined that the company will not make the mistake it did in the late 1970s, when it forecast that personal computers would be a low-volume hobbyist market.

'In making that prediction, we talked to people in the computer market as it then was - essentially mainframe manufacturers,' he says. 'Today we are talking to an extremely diverse group of people.'

Those discussions have led the company in two directions. The first is the development of aPC add-on that allows different computer users to see each other via TV links while they work jointly on spreadsheets and documents. Early versions come out next year.

'One business journey avoided and you've paid for the equipment,' Mr Grove enthuses.

The second plank of the strategy is to find non-PC markets for PC chips. By employing the company's 386 chip, the predecessor to the 486, as an 'embedded controller' in the growing number of intelligent domestic and office devices, the company avoids R&D expenditure and can continue to wring revenue out of expensive factories. The market could dwarf the PC business, and the company has formed an association with Microsoft to develop intelligent televisions.

Mr Grove, the author of two management text-books and a part-time lecturer at Stanford Business School, is well aware that these ventures take Intel far outside its core competencies into tougher markets with keener prices and lower costs of entry.

'Intel's heroes are risk-takers who made things happen,' he reflects. Whether his own place alongside earlier heroes in the Intel museum is assured remains to be seen.

(Photograph omitted)