Will wireless drive the semiconductor industry?

By Bob Mueller -- Purchasing, 12/22/2000

At the beginning of last year, the top four consumers of semiconductors were Compaq, IBM, Dell and Hewlett-Packard, all of them PC heavyweights. But trailing not far behind them, according to a study by Gartner Dataquest, were Siemens, Ericsson, Motorola and Nokia, all best known for their cellular phones.

Are cell phones and, more broadly, wireless devices in general, about to edge out PCs as the driving force in the semiconductor market?

The short answer is, probably not anytime soon. It's true that demand for the silicon used in cellular handsets and other wireless devices is the semiconductor industry's biggest growth area and some components are in short supply. But Gartner Dataquest numbers quoted by Advanced Micro Devices (AMD) in a recent presentation to analysts project that by 2003, communication applications will account for less than a quarter of the semiconductor industry's sales, compared to nearly half for computers.

The market for cellular handsets, by far the largest component of the wireless market, has grown at a rate of around 50% annually for the past several years, according to Gartner Dataquest Chief Analyst Stan Bruederle. In 1999, the industry sold 290 million units worldwide. Last year, that number grew to an estimated 420 million. "When you get a market that's that big and is growing that fast, it's going to put a lot of stress on all the supplier categories and, in fact, that's what's been happening," says Bruederle.

The biggest squeeze right now is in flash memory, non-volatile memory chips used in cellular handsets to store phone numbers and-as more handsets become Internet-enabled-e-mail and other data from the Net. The same technology is used in a variety of other wireless products, in automobile engine control and entertainment systems, in consumer electronics products such as MP3 players and digital cameras, and in PCs to store the BIOS and power-up code.

Tight supplies

Both Intel and Fujitsu AMD Semiconductor (a joint venture of AMD and Fujitsu), the two biggest producers of flash memory, are contracted for their entire production through 2001, and some analysts see supplies running tight well beyond that. So far, cell phone buyers aren't seeing any handset shortages, however, says Bruederle. "People are still able to get phones when they need them. There may be some localized challenges, but I would say overall the problem doesn't seem so severe."

But it's possible shortages could constrain growth in the future. Some cellular industry observers have trimmed their estimates of future sales for that reason, and although contract pricing for flash memory is stable, prices on the spot market are running several times higher, says Walid Maghribi, AMD group vice president, Memory Group.

The big demand in flash memory is for high-capacity chips that tend to end up in communication devices, says Tony Sica, marketing manager for Intel's Wireless Communication and Computing Group. "As you start getting into the higher-end phones and as you start seeing phones that can browse the Internet, the density in the demand for flash bits goes up pretty dramatically. That's where our products like StratoFlash and the large-block flash devices that we produce go."

Flash memory requirements for PCs, on the other hand, are modest, and beyond the wafer level, PC flash doesn't contend with wireless flash for production capacity. The same goes for processor chips, Sica notes. Both cell phones and PCs use them, but the designs for each are quite different. In cellular applications, low power consumption is a primary objective. In desktop PC applications, performance is what matters most. Power availability is assumed to be inexhaustible. Moreover, demand for processors in wireless today isn't especially great, Sica continues. "But in three to five years, you could see that story changing as we start seeing more and more of these Internet devices, or Internet clients, hit the marketplace."

In the complex world of semiconductor manufacturing, the drive among companies like Intel to increase PC processor performance may actually wind up improving the supply of other chip products like flash memory, says Bruederle. Switching manufacturing in fabrication plants from one type of semiconductor to another is not a big deal. "As Intel and AMD move from one generation of processor to the next, they require more advanced fabrication facilities," Bruederle says. "So they'll take the facility they were using to manufacture microprocessors, and they'll shift it into production for, say, flash memory."

Production capacity

Adding production capacity in the memory business is always dicey, says Sica. "It has its own biorhythms and its own market ups and downs," he says. "You can be constrained today, and three quarters from today you could be at a low point. We see it in dram all the time."

Nevertheless, Intel is bringing a new flash memory fab online in first quarter 2001 and adding capacity at some existing plants. John Greenagel, director of corporate communications for AMD, notes his firm will have a new flash fab up and running by midyear. "We can't add capacity fast enough to keep up with demand," he says, noting that even the manufacturing equipment is back-ordered.

The Internet is a big driver of both PC sales and wireless technology. Could wireless devices become the principle way users gain access to the Internet in the future, and might that help those devices overtake PC sales? Sica thinks not. "We don't see a significant decline in the growth of the desktop PC market," he says. "One of the things you have to look at is that the best Internet experience you can have today happens on a PC. People will evangelize other solutions, like Web TV, but when you get down to the experience of playing Internet games, getting good, rich multimedia content, it's the PC today."

Even if there were a way to make cell phone screens bigger or if handhelds like Compaq's iPAQ catch on, Internet performance still won't equal what PCs deliver, Sica argues. "My position is, we'll see growth in the [wireless Internet access] market over the next three to five years as the infrastructure becomes able to support that kind of rich multimedia on the phone. More than likely, it will not damage PC sales."