Analysts clash over chip industry forecasts
pubs.cmpnet.com
AIX-LES-BAINS, France - Though friends, the respective European
analysts for Dataquest Inc. and World Semiconductor Trade Statistics
(WSTS) appear to clash on the directions the semiconductor industry
should take in light of currently declining orders and falling prices, which spell a gloomy industry forecast for 1998 and '99.
A dramatic decrease in capital spending is the only cure for the
overcapacity that has depressed prices and industry revenue, according to Jean-Philippe Dauvin, president of the European branch of the WSTS and economist for SGS-Thomson Microelectronics.
But Dataquest's European business analyst held to an optimistic forecast. "The industry will double in the next five years, in spite of what you're learning now," said Joe Grenier. "If the industry is to accommodate the $288 billion market [projected for 2002], the industry must build 35 new fabs per year."
But Dauvin said he is worried about semiconductor industry performance. Industry bookings appeared to drop precipitously in January and February of 1997, then gradually improved through the end of the year. That led the WSTS to suggest a "gradual market recovery" in 1998, with perhaps 6 percent growth for the year, up from the industry's modest 4 percent growth in 1997. But at the book-to-bill ratio "turned south" at the beginning of 1998, Dauvin said, and in consequence all performance estimates for the year must be revised downward.
This situation is compounded by the effects of the Asian currency crisis, such as very soft Asian demand, a slowing of the Japanese supply chains into Asia, and a virtual halt of U.S. PC production, Dauvin said. His semiconductor forecast [remarkably similar to a forecast presented at a Texas Instruments analysts meeting in March] shows the strongest growth in semiconductor consumption this year coming from Europe (at 11 percent), followed by the United States (with 9 percent projected growth). The Asia-Pacific region will turn in a modest 8 percent growth, while Japan's market will decrease 6 percent. Thus, worldwide consumption will increase only 6 percent.
Dauvin expressed serious concern over spending on capital equipment,
especially on investments in new manufacturing facilities. Capital
expenditures decreased 10 percent to $40 billion in 1997, and - in light of current conditions - are projected to decrease by 14 percent to about $34 billion in both 1998 and 1999. But this cloud has a silver lining, Dauvin said. Decreases in capital spending, especially in Asia, will decrease the overcapacity that has depressed prices and revenues. Thus, average selling prices and overall revenue will be healthy for newer high-margin products that are expected to go into volume production in 2000, such as chips for mobile phone, digital consumer goods, smart cards and automotive gear.
Happy face
But Dataquest's Grenier described a near-term outlook that sounded more optimistic. Given the $910 billion of electronic products - with $147 billion in semiconductor content - sold in 1997, he sees total electronic products growing to $1.25 trillion in five years. The three big drivers that will account for 70 percent of all semiconductor activity are PCs, communications and consumer electronics, he said. Consumer ICs are projected by Dataquest to be a $40 billion market by 2002. And communication ICs are project to grow from the $32 billion in 1997 to $60 billion in 1998. Advances in communications ICs, especially LAN and WAN chips, are now driving communication applications - rather than the other way around, he said.
DVD chips will see 100 percent growth from 1996 to 2001, while xDSL
devices see 60 percent growth and automotive GPS chips will have 55
percent growth over the same period. But DRAM revenues will be halved
from $40 billion in 1997 to $21 billion in 1998. The DRAM industry will ship more product, he said, but will get much less money for it.
Contradicting Dauvin's belief that the PC market is saturated, Grenier anticipates further growth. PC units will continue to grow 16 percent per year, but depressed priced will lead to only 8.9 percent revenue growth per year. Intel's overall revenue, he said, will likely be flat. Grenier agreed on this point. Refering to the margins on a $250 CPU, Grenier concluded: "That era is over now."
The analysts seemed to agree on the long-term outlook for the industry. From 1967 to 1997, Dauvin said, the industry showed compound annual growth of 15 percent.
That figure is not likely to change, he said. "This business can go up and down daily," he concluded, "but this business will be 15 percent forever." |
