Intel Profile: Silicon Hymn Plants

It?s easy to sing the praises of a company which doubles its chip performance every year or so. But how does Intel plan to equal this feat in the future? Paul Bray investigates

?God was in a good mood when he created silicon,? said Intel CEO and president Andy Grove at a customer conference in London last month.

The Almighty has certainly smiled on Intel. His philanthropic creation has helped the company to become one of America?s most successful and most admired businesses, with each employee earning the company nearly half a million dollars. Last year, turnover rose 29 per cent to $20.8 billion and gross margins topped 60 per cent.

Since inventing the microprocessor in 1971 and specialising in the production of microprocessors since the mid 80s, the company has doubled the performance of its chips every 12 to 18 months. It expects to go on doing so for at least the next 15 years. By that time, says Grove, Intel processors could hold a billion transistors and run at 10GHz (a Pentium Pro has 5.5 million transistors and crawls along at just 200MHz).

Intel chairman and co-founder Gordon Moore put it like this: ?If the auto industry advanced as rapidly as the semiconductor industry, a Rolls-Royce would get half a million miles per gallon, and it would be cheaper to throw away than to park.?

So is Grove happy? Up to a point, perhaps. But he is famous for believing that only the paranoid survive, and thinks that technological advances on the scale outlined are not just possible, but essential if Intel is to keep itself and its shareholders in the style to which they have become accustomed.

Intel is driven by competition. ?Competitors have an awful lot of effect on us,? says Steve Poole, formerly Intel?s co-general manager for Europe and now director of European operations. ?We only stay ahead because we drive technology forward at breakneck speed.?

Competitors such as AMD and Cyrix have only a small market share, but their ability to clone the Intel range up to at least the previous generation of chip means Intel must always be coming up with something better to maintain its high profit margins.

But this is only half the story. In 1995, Intel announced that it would be building a second fabrication plant, or ?fab?, at its European manufacturing base at Leixlip in Ireland. It will cost the company $1.5 billion, yet the first chip will not roll off the production line until 1998.

The cost and time-lag are typical for Intel fabs, of which the new Leixlip plant will be number 14. With up to $2 billion committed at least two years in advance for each fab, Intel has to be sure it can sell all the output. Small wonder that Grove calls the fabs ?fields of dreams?.

Until about 1990, Intel felt it could rely on the PC industry to create enough demand. As each new generation of chip appeared, there was always a PC manufacturer ? usually Compaq ? eager to rush it to market, and a bunch of corporate power-users willing to pay the price to be able to recalculate their spreadsheets faster or redraw their diagrams better.

Now things have changed. Competition is cutthroat and PC margins are small, making manufacturers reluctant to innovate. ?We can no longer depend on PC manufacturers for the technical improvement of the PC platform, because the market is too intensely competitive,? says Poole. ?Platform improvements of any significance are very expensive to design and fairly expensive to get customers to adopt.

?That?s why we set up the Intel Architecture Labs, which at the peak had almost 1,000 engineers, devoted not to designing chips or products, but to figuring out what improvements should be made to the platform capability, and then working with the software and hardware industry to get those improvements standardised and accepted.?

The labs? biggest success was the design and introduction of the PCI bus, intended to reduce the bottleneck between the processor and other components, and now an industry standard. Another was the video standard, DVI.

Now Intel is pushing for new standards in networking, specifically in the management of networked PCs. It has taken on board the criticisms of the Gartner Group and others, which have found that the cost of maintaining and supporting a PC is several times the initial purchase price. These have been borne out by Intel?s own experience, where it found its own 50,000 or so PCs were costing it more to run than the industry average.

In 1992, Intel set up a desktop management taskforce of about 200 hardware and software firms, including Compaq, Hewlett Packard, Novell, SCO and Symantec. This has now devised a set of specifications called Wired for Management, many of which are built into the chipsets and motherboards of Pentium Pro PCs. Intel is also working on both network infrastructure and high-bandwidth projects.

The Wired for Management features will not be fully retro-fittable to standard Pentium machines, no doubt giving corporates an extra incentive to move their base standard up to the Pentium Pro.

Intel needs to boost demand for Pentium Pro if it is to maintain the breakneck momentum which will keep it ahead of its competitors and ensure continued high profits. The company admits it was surprised at the slow initial take-up of Pentium Pro, presumably caused by the lack of availability of the 32-bit software necessary to exploit its full performance.

The company expects demand to increase this year. ?We think corporate accounts will probably stop buying standard Pentium machines later this year, and increasingly focus their purchasing on Pentium Pro,? says Poole.

Intel makes no secret of its willingness to manipulate chip prices to push a particular generation of technology, so this could be an indication that Pentium Pro prices are about to be cut to commodity levels.

The company has already set the price of the new Pentium MMX chips almost identical to that of the standard Pentiums they are intended to replace, thereby making their success almost a foregone conclusion.

MMX marks a break with tradition for Intel. It is the first time since the development of the 386 that the company has added to the instruction set (the 486, Pentium and Pentium Pro were all designed to execute the existing instructions faster and more efficiently).

Intel dismisses suggestions that piling more and more capability into the processor, its associated chipsets and Intel-built motherboards will stifle competition among PC and component manufacturers by reducing the scope for innovation. But with functions such as network management, video decompression and 3D rendering taking place within the chip, rather than needing separate components, this could well be the effect.

Grove?s customer presentation included a prediction that increasing use of the internet will put customers directly in touch with manufacturers and push out the middlemen in the distribution and telesales industries. Ironically, he did not make the connection with what his own company is doing in the PC industry. No wonder Intel says it can no longer depend on PC makers for technological improvements.

But merely taking control of the PC?s development will not be enough to ensure Intel?s continued success. PCs already have ample power for personal productivity tasks, and the more powerful they become, the more reluctant their users are to junk them and replace them with something faster.

As well as developing the technological infrastructure to support its increasingly powerful processors, Intel needs to persuade customers that they need to buy them.

?If we don?t make computers more useful, there won?t be demand for the chips we?ll be making in a few years,? Grove told Fortune magazine. ?So we have to create users and uses for our microprocessors. We get the market growth we earn by our development efforts, our investments and our proselytisation.?

In the past, the demos at Intel presentations showed the latest power PC displaying fractal images or running Mandelbrot tests ? about as interesting as a demonstration of quicker-drying paint. Grove?s presentation last month featured a live link-up with BMW in Germany via a Pro Share video conferencing link.

This is the kind of application in which Intel must interest customers if it is to continue selling its ever-more powerful chips. An internet phone and internet videophone are already in beta, and Intel?s labs have been busy turning out technology for real-time internet video as well as standards for internet telephony.

Intel has invested over $500 million in multimedia companies which need or will need high-performance computers, such as Web broadcasting specialist Cnet or 3D conferencing company Oz Interactive. It has increasing interests in the entertainment industry, such as in the Creative Artists? Agency Media Lab, which aims to provide multimedia entertainment content for PCs.

Consumer sales will be crucial to Intel?s success. The company?s vision is of a PC which will replace the television and hi-fi as the main home entertainment centre. At Comdex last year, Grove said: ?We need to look at our business as more than simply the building and selling of personal computers. Our business is the delivery of information and lifelike interactive experiences.? He has even coined a new catchphrase ? the war of the eyeball ? to define the struggle for householders? hearts and minds.

On the biggest potential stumbling block for consumer sales ? the fact that their new PCs may be obsolete in two years ? Intel is bullish. Poole says: ?All we can say to people is that you need to spend as much money as you possibly can. But you have to accept that the machine you?re buying is probably going to be outmoded in two to three years. It isn?t something we feel guilty about.?

For the sake of everyone in the PC industry, Intel had better be right. ?It?s because the PC is very different from other consumer goods that we see the ongoing growth of volume in the PC business,? says Poole.

?It?s not just repeat business because it?s worn out, it?s because there?s new capability being introduced all the time. If the industry started to lose the knack of doing that, the nature of the PC market would change dramatically, and it would look far more like the mature volume consumer markets like TVs and videos.? Which would mean less work for all of us.

For all its undoubted ability in technological innovation, Intel?s main objective must be to maintain its existing momentum, and the faster it runs, the longer the open road it needs to be able to see ahead. For Intel, the great inventor, the greatest challenge is in continually reinventing itself.