Selling the Souped-Up Chips
By Peter Burrows
Truth in marketing may be a squishy concept from the consumer's point of view, but it's pretty simple from the marketer's point of view: Sell what you got. For decades, chipmakers have been able to offer speed. Whether measured in megahertz or, as in recent years, in gigahertz, processor-makers such as IBM (IBM ) and Intel (INTC ) have been able to crank up the speed of their wares with amazing predictability -- and get consumers to spend money on them.
Until now, that is. In the past year, the fastest chips increased just a pittance compared to past years -- from 3.2 ghz to 3.6 ghz, in the case of Intel. Chipmakers have been struggling mightily to get back on pace with historic trends. The problem is, making chips run faster also makes them run hotter. Not only could that result in uncomfortably warm laps for notebook users, more of the chips burn up in production, lowering manufacturing yields.
The evidence abounds. In recent months, Intel and Sun Microsystems (SUNW ) have nixed development of upcoming speedsters, after becoming convinced that the gigahertz improvements wouldn't justify the investments. IBM has been able to build so few of its new G5 microprocessors that Apple Computer (AAPL ) recently delayed the introduction of a new iMac model from July to September. And on July 30, Intel announced that it would not introduce a new 4-GHz Pentium until next year, rather than in the fourth quarter.
WHO MENTIONED FAST?
"They're still having trouble building enough of the 3.6-GHz chips," says Kevin Krewell, editor-in-chief of the Microprocessor Report, a trade magazine. Indeed, PC makers may not get 5-GHz processors to hawk in their systems until 2007 or later, says chip analyst Nathan Brookwood of Insight64.
So, what's a chip marketer to do? Tell a different story that focuses on something other than speed. That process has already started. In March, Intel changed the way it names its chips; they no longer include references to clock-speed. Advanced Micro Devices (AMD ) has been pushing its True Performance Initiative for a few years, in which consumers are urged to focus on buying the chip that will best run the application they most intend to use.
Look for this remarketing of microprocessors to pick up steam. What will the newest thrust be? Given all the viruses wreaking havoc with consumers and businesses, security could be one major theme. Many chips due out later in 2004 will be able to exploit a new Microsoft (MSFT ) technology called NX that will help PCs avoid certain viruses.
But the biggest marketing thrust will be toward so-called "multicore" processors. The basic idea: Rather than use the ever-increasing number of transistors that can fit on a slice of silicon to build one red-hot, electricity-guzzling 4-GHz screamer, why not lay down two cooler-running 2-GHz cores instead? Says one PC company CEO, "The multicore is a reaction to the fact that climbing the gigahertz ladder has come to a halt. If you want to sell more microprocessors, you have to do something new. If you can't go high, you have to go wide. Chipmakers don't have a choice."
For the uninitiated, there's a fair amount of tech two-stepping in this marketing approach. But multicore chips will, in fact, improve PC performance, especially as people use their PCs more as entertainment devices. Let's say you're using your PC to record your favorite TV show, and talk to your favorite aunt over a voice-over-Internet-protocol phone at the same time. If your PC has a dual-core processor, each core could handle one of these jobs -- rather than force a single-core chip to try to jump between the two tasks without dropping images or words. Says Krewell, "By having two cores, you could get 50% to 80% better performance on multiprocessor-aware software."
These chips are coming to market fairly soon. As of now, Intel and AMD are expected to unveil dual-core processors in late 2005. IBM is on a similar timetable, with a chip architecture called Cell that will be in computers and other products, including Sony's (SNE ) next-generation PlayStation. Sun Microsystems is pursuing the most radical multicore strategy, starting with chips that will have eight cores on one piece of silicon.
"LIKE THE CAR INDUSTRY."
Multicore chips won't be a marketing panacea for chipmakers. Single-core chips will probably continue to be in the majority of tech products for the next few years. Going forward, trying to convince consumers they need that 32-core chip rather than their current 2-core clunker won't be easy -- certainly not as easy as hawking sheer speed has been until now. In a society that has always understood the power and allure of speed, millions of people have forked over a few hundred extra dollars for a PC with a few more megahertz, even though it was widely known that the incremental oomph had little real impact on PC performance.
That means over the long haul, PC marketers will have to hone their multicore message and find better ways to highlight the real advantages. "It's like the car industry," says analyst Brookwood. "Years ago, cars were sold based on horsepower. Today, most people are more concerned with how many cup-holders the car has, or whether it has a DVD player or GPS system." It's just one more sign of how the tech industry is moving into a new phase of evolution -- and figuring out how to bring consumers along.
Burrows is Computer Editor in BusinessWeek's Silicon Valley bureau
Edited by Beth Belton