At the end of 1999, when Intel became a component of the Dow Jones Industrial Average, I chastised that financial reporting company for reducing the value of this once-respected measure of economic health. "Exxons oil in the ground and refineries above the ground will hold their value far longer than Intels soon-to-be-obsolete chip fabs," I argued.
Intel, with its high-margin business model, which compels it to reinvent its entire technology base every five years, simply cannot be compared to an ExxonMobil, an Alcoa or a Caterpillar Tractor (all of them respectable industrials) as an economic indicator.
Intels transition to Itanium technology renews my concerns. As we began this month, eWeek reported on Compaq customers reactions to the companys shift from the proven Alpha design to the still-nascent Intel Itanium as its foundation for future 64-bit systems. One customer, the chief technology officer at NASDAQ, was quoted as saying, "Im comfortable with Intel. All of our Web sites [and] networks run on Wintel." And his point is?
Perhaps Itaniums upward-compatible path from the X86 instruction set will appeal to moderate-intensity Wintel users, but whats the appeal for enterprise sites (including NASDAQ) that are migrating mission-critical, 24-by-7 duties from their MIPS-based Himalaya machines? How many risks do these sites really want to take at once? New hardware, new compilers, new operating systems: As Larry Niven and Jerry Pournelle once wrote, "We juggle priceless eggs in variable gravity."
According to one Wall Street analyst, its safe to predict that the chip-making business has nearly hit bottom—because, said ABN Amros Paul Leming, "if it doesnt hit bottom in the next month or two, itll cease to exist." Amusing, but rather missing the point. People need chips, just as they need oil, aluminum and earth movers: as inputs to other industry. "Architecture du jour" is not the path to recovery.