SAN MATEO, Calif. — Market stability has always been an elusive goal for U.S. chip vendors, but that hasn't stopped the experts from predicting the end of the boom/bust cycles every so often. In the mid-'90s some prognosticators thought feast-or-famine days would give way to steady, uninterrupted growth as the PC boom hit its stride. But these hopes were soon dashed when it became clear that chip makers were as vulnerable to price shifts in DRAMs as pig farmers are to the price of pork bellies.
Now, looking to rebound after the worst chip recession on record, chip makers and crystal ball-bearers are giving it another go. Sales, they say, should grow moderately from year to year as chip makers get used to the idea that most of the end applications — and the rate at which they absorb new technology — are pretty well known. This is another way of saying that the chip industry, in the United States at least, has reached middle age.
The U.S.-based Semiconductor Industry Association, for one, seems to have embraced this outlook, though perhaps belatedly. Last October the SIA predicted that chip sales would grow by 19 percent this year, the kind of bullish forecast that reflects the industry's traditional bipolar temperament. In recent weeks, however, the SIA brought it down to 10.1 percent, and other analysts too have moderated their growth predictions. One reason is that hopes for higher rates of PC consumption haven't materialized. The SIA blamed other culprits too, like the war in Iraq and the SARS outbreak in Asia.
By 2004, the SIA thinks growth should jump to 16.8 percent, though this will be fueled largely by a long-overdue uptick in sales of memory chips, the most volatile of chip categories, which is served by a short list of vendors these days and is hardly representative of the industry as a whole. Indeed, the SIA believes these gains will be short-lived, and by 2005 and 2006 growth should be less than 10 percent.
"We end up with a compound annual growth rate of 10 percent, which is consistent with what the market will demand going forward as a maturing industry," said SIA president George Scalise.
The industry is showing other signs of age. In earlier years chip companies would plow profits right back into product development so as not to miss the next big thing. But companies like Intel, Linear Technology, Maxim, Microchip and Texas Instruments have started paying out quarterly dividends. Intel chief financial officer Andy Bryant recently explained this cash-is-king mantra to investors at a Bear Stearns conference. "What matters to us and what matters to investors is how much cash you're going to generate from your business and what share you get as an owner," he said. "In the end what counts is making returns and turning it into cash."
Yet this shouldn't imply that U.S. semiconductor companies can afford to be complacent. One of the challenges that chip makers here are facing is how to compete now that the center of gravity in the electronics industry has shifted to Asia. Today, nearly $2 in semiconductors is sold in Asia for every $1 sold in the United States. In fact, the Americas is the only region expected to experience a decline this year while Europe, Japan and Asia all enjoy double-digit growth rates in chip consumption.
Asia's high consumption rate is not unexpected. What's less obvious is that much of the design and engineering work for networking gear and cellular phones is making its way across the Pacific.
"The engineering decisions are being made in Taiwan and now even in China," said Jim Jones, director of BA Venture Partners, a division of Bank of America. "If chip makers don't know the right price point to hit they'll have a hard time."
As more of the contract manufacturing and engineering moved to Asia, so too has much of the chip-manufacturing capacity. Chip makers in the United States can learn to manage the shift of design and manufacturing of equipment to Asia so long as they can build up the sales and marketing presence in local markets. But if too many U.S. chip companies stop investing in chip manufacturing at home, the industry risks losing its technology edge.
Still, fewer U.S. chip companies can afford to invest in leading-edge manufacturing capacity. The shift from 200- to 300-mm wafers is one change that will separate the haves from the have-nots. The move from 130- to 90-nanometer process design rules is another.
Intel's Bryant said only those companies making $6 billion a year in revenue will be able to do it. By that measure alone, that means Intel and Texas Instruments in the United States. IBM Microelectronics and Micron are still going down that path but their revenue in 2002 was less than $3 billion.
The mid-tier chip companies will have to find other ways to control costs without abandoning manufacturing altogether. Some, such as Motorola, have adopted a "fab-lite" strategy, closing some fabs at home and shifting more manufacturing abroad. Others are using the capacity they have left or they're shutting lines down completely and shifting all their manufacturing to Taiwan. That's led some chip makers, including Atmel, Microchip, Vitesse and Zilog, to put fabs on the block.
There's also a glut of capacity in the United States, especially older lines that process wafers less than 200 mm in diameter. "There's a ton of 4- and 6-inch lines that have been mothballed and just sitting on the sidelines," said Stephen Rothrock, managing partner of the microelectronics division of real estate firm Colliers International.
U.S. chip makers have clear reasons to stay on top of manufacturing, which is why organizations like SIA pump some $65 million a year into research, sharing the fruits of that labor with industry. Local and state governments too should devise policies "that will help manufacturing become more competitive in their locale," Scalise said.
