How Intel spreads its risks…

Overegging the chip omelette

Column The day was, and it wasn't that long ago, that Intel was perceived as a one product company. That product was the x.86 microprocessor and we noted some four or five years back how it was extending its tentacles into other areas of the PC industry, including, or especially, the networking business. From a position of being nowhere, last year Intel became the top worldwide supplier of network interface cards (NICs), toppling 3Com from its commodity pile. How prescient Intel was. Speaking at its analyst meeting last week, CEO Craig Barrett presented Intel's mission as now being the "building block company", a phrase we first heard at its Developer Forum in February but which it has repeated, at regular occasions, whenever it's had the chance, since. Almost unnoticed, Intel has positioned itself in a strong position by acquisition -- practically one a month this year. And at the last Developer Forum in September, it took the unprecedented step of spending quite a bit of time telling the world's press about the reasons behind its acquisition of key network technology, rather than just banging on about its latest chips. A quick look back at Intel history reveals, however, that the company has always realised its vulnerability, depending on who the contenders were and what the marchitects perceived as the threat. Intel doesn't always get it right, it has to be noted. Barrett and the other Intel execs spent some little time at the conference call last week promoting the virtues of the forthcoming Timna chip, a useful little number with an x.86 core which is really a system on a chip and is intended to be a thorn in the side of National Semiconductor and its Geode. Brian Halla, CEO of NatSemi, realised around this time last year that competing head-to-head against Intel with its Cyrix x.86 offerings was a waste of time and money, and spent his time pushing a system-on-a-chip concept, hurrying up his designers across the world and so able to introduce the Geode in August. At the time, Halla and other senior executives at NatSemi made it clear that there was only a limited window of opportunity and so have ramped up Geode I in volume, while designing Geode II at the same time. In price and performance terms, NatSemi has the advantage that the Geode exists, in volume, and the company has already had a number of design wins, as we have reported. NatSemi has a fairly respectable track record in embedded systems, while Intel has often seemed to regard this market as a place to sell chips it has already made redundant. There isn't a great deal of margin in chips for set top boxes and appliances, however. NatSemi refused to tell us how much the Geode cost but we're not talking Coppermine talk here. Intel, meanwhile, has had to face the fact that its high end and lucrative Pentium desktop market, and even its even more high end Xeon server chip line has suffered price erosion during the year, because of competition from AMD and because time, in the shape of other companies including IBM, Sun and Compaq, is ever merciless. The Internet market is a very different kettle of fish and by positioning itself as a major building block player, Intel has something of an advantage on its competitors, notably IBM. Big Blue has the stretch and technology to take on Intel and beat it but while it has the architectural strength, it doesn't have the marchitectural infrastructure or Chipzella's worldwide reach. So Intel will continue to turn in good results on its microprocessors, fend off competition from AMD on the x.86 front, and position itself as a real power in the system-on-a-chip market. While the world waits for Coppermine chips in volume, Intel will continue to supply .25 micron parts, profitably and through its well-established sales channel. But Chipzilla being all things to all men in all of the industry? We'd rather doubt whether even Intel could pull that one off. ®

Sponsored: The Nuts and Bolts of Ransomware in 2016