Issue



ISS97 is less bubbly; Outlook cloudy until 98


03/01/1997







ISS `97 is less bubbly; outlook cloudy until `98

Peter N. Dunn, WaferNews

January`s Industry Strategy Symposium, sponsored by SEMI, once again drew several hundred industry executives and observers to Pebble Beach, CA, for a look at market conditions and outlook (plus a few rounds of golf).

The gathering was less bubbly than the 1996 edition, which bordered on euphoria; the effects of last year`s semiconductor and equipment market downturns did not hit until a few months after ISS `96`s happy predictions of the end of market cycles and uninterrupted growth (see "Strategy Symposium Snapshots," SST, March 1996, p. 50). Much of the 1997 discussion centered on how long and deep the current downturn will be, and what factors will drive the next upturn when it comes. Another highlight was a discussion among stock market analysts who follow the equipment and materials sectors, who shared insights into how they go about selecting the companies they recommend to their clients.

A number of presenters from the market research community took time to explain part forecasting miscues to the audience of top executives, and growth estimates were toned down a bit compared to last year`s numbers, but there was still a broad consensus that chipmakers and their equipment and materials suppliers have some very good years ahead.

Equipment companies, however, will not have one of those years in 1997. While opinions differed on how long the current downturn will last, with VLSI Research staking out a relatively optimistic position and Dataquest a more bearish outlook, there was agreement that the equipment market will shrink by about 15-18 percent this year. VLSIR president Dan Hutcheson predicted that 1998 sales would be slightly higher than 1996`s, while Clark Fuhs, principal analyst for Dataquest, said 1998 will be about flat with 1997, before an uptick in 1999 (see figure).

"People want good news about the second half of `97, but we just couldn`t do it," said Fuhs. He suggested that equipment firms are now holding a four-month backlog of orders or less, compared to the six-month level in place a year ago. He predicted that toolmakers would continue to ship at 4Q96 rates in order to keep their financial numbers healthy, but said "disappointing orders in the second half" would result in a drop in 3Q shipments. Real growth is not seen until the latter half

of 1998 and 1999, when DRAM capacity adds begin to kick in.

Fuhs added that "fab capacity coming on in 1997 exceeds the incremental demand," just as it did in 1996, when supply increased about 14 percent and demand only 4 percent. On the upside, Dataquest sees "no major hiccups" in PC market unit growth of 16-18 percent annually; PCs account for about a third of semiconductor sales, and are a crucial driver for future capacity expansion.

Hutcheson sounded a somewhat more optimistic note, stating that fab utilization is actually higher than generally thought. In 4Q96, he said, utilization began to tighten for the first time since 1995, with December utilization coming in at 98.6 percent worldwide. "There`s the potential for an upside surprise in 1997," he said. "Equipment sales will be driven by the move to 0.25 micron from 0.35 micron. If chip companies want to grow, they need to invest." In later comments, he added, "From the perspective of running a business, `96 was the worst year. The issue was managing cancellations. In 1997, we will probably see booking strength by 3Q, with billings rising in 3Q and 4Q."

Bill McClean, VP for market research at Integrated Circuit Engineering Corp., said the overall semiconductor market should show 11 percent growth in 1997, with growth accelerating through 2001, when it is predicted to hit 24 percent. He agreed with Texas Instruments` chief economist Vladi Catto, who predicted a $300 billion semiconductor industry in 2000, but said it might be 2001, instead. "We see excess DRAM capacity through 1997, and a relative balance in 1998," said McClean. Looking at capital spending, McClean stated, "The Japanese companies have turned back; the Korean companies are spending for five years from now, as the Japanese did in the mid-80s." He noted that LG Semicon spent an estimated 104 percent of semiconductor revenues on capital investment and noted, "They`re not making any money." Taiwanese firms, he continued, "are keeping up a brave front, but I don`t know how long it will last, with the number of companies who are talking about getting into DRAMs."

Looking to the materials side, market researcher Dan Rose, president of Los Altos, CA-based Rose Associates, said the semiconductor materials industry is poised for several years of solid growth, but the advent of new process chemicals and compounds is likely to spawn many integration challenges as design rules drop below 0.25 micron. He noted that the semiconductor industry downturn of 1996 had actually been good for the materials sector, providing some relief in the tight supplies of polysilicon, prime 200-mm wafers, fluorine-containing gases, and quartz. However, supplies of epi wafers and the most advanced photomasks are still constrained, and steady unit volumes helped the materials industry grow 9.9 percent in 1996 even as semiconductor sales dipped.

For the fab materials market as a whole, Rose estimated 9.9 percent growth in 1997. Annual increases of 12.0 and 12.9 percent are seen in subsequent years.

Polysilicon producers are hastening to expand their plants, noted Rose, with an eye toward doubling 1995 capacity in the 1999/2000 time frame. Hemlock, MEMC, and Mitsubishi Materials are all bringing major new facilities on line by 1998. "Without the downturn, there would have been a 2000-ton shortfall in poly -the downturn saved the day," commented Rose. He suggested, "the Japanese will probably lead the charge in 300-mm adoption; pay attention to their forecasts."

The rate of new materials adoption, said Rose, is "a looming problem," as ever-tighter geometries force a proliferation of new compounds in the fab. He noted that in the early days of the industry, fewer than 10 separate materials were needed. In the submicron era, about one new material has been added per generation (Ti, Cu doping, W, etc.), but as many as four might be needed both at 0.18 and 0.13 to cope with new metallization and dielectric needs.

"Sequential adoption of one material at a time allows for good process engineering," commented Rose. "You can hold the process stable and add one variable. But at 0.18 and 0.13, you will have to do a matrix instead of holding stable."

Other points from Rose`s talk:

 Dilute alternatives to the RCA wafer clean are now proven, and will result in a decline in the traditional wet chemical market after the turn of the century. "Fujitsu and NEC are committed to using dilute cleans by 1999," said Rose.

 Multilayer laminate packages are arriving in force, initially in the automotive and cellular sectors and soon in high-speed logic at companies like Intel. "The transition brings the PC board industry face to face with the semiconductor industry," said Rose. "The PCB people aren`t used to working with fine-line geometries, and there are opportunities for startups. I expect it to be a billion-dollar industry in the next three years."

 Materials suppliers are facing "incredible complexity" as chipmakers outsource both wafer fab and packaging and assembly operations. Specifications, pricing and delivery plans, and field support issues multiply, even as IC companies seek seamless integration. "Global suppliers are taking share from regional suppliers," said Rose. "They`re setting up global networks, but that increases overhead and results in increased materials costs."

At the "Bulls and Bears" panel, four analysts discussed their selection strategies and other stock market issues with moderator Jerry Taylor, senior VP and CFO of Applied Materials.

All the analysts agreed that management issues are at or near the top of their checklists when picking companies to watch. "The key word in a management team is `team`," said Elliot Rogers, managing director of Deutsche Morgan Grenfell. "I don`t like one-man shows. Do they see the big world? Do they have expertise in the areas of R & D, marketing, and manufacturing?"

"The quality of management is the most important thing," added Brett Hodess, managing director at Montgomery Securities. "Also the growth capability, which is driven by technology, the competitive landscape, and the company`s track record of product transitions."

Dan Kleskin, managing director at Robertson, Stephens & Co., said, "I look at experience -do they have 5-10 years, or 15-20? How broad is the experience -is it just the CEO, or is there a whole team of experienced people? How well do they work together and communicate? Is there a good CFO - the CEO and CFO are the two most important contacts to Wall Street."

Hodess added, "The execution of the management team is important. Have they been able to make product transitions when they say? Some companies introduce a product and have no business for three years, others release with a backlog. Also, you have to have execution on the manufacturing side."

Kleskin added that he also likes to see "a defensible, unfair advantage in terms of technology or a patent position -an ability to put up barriers. Good-quality management usually has a good product line and an advantage."

Two views of the worldwide fab equipment market, from Dataquest (top), and VLSI Research (bottom). Overall numbers differ because of different counting methods, but the shape of the curves shows the divergence of opinion.

Peter N. Dunn is editor and publisher of WaferNews, the only weekly news publication dedicated to the semiconductor manufacturing community. For more information, contact Dunn at ph 603/891-9121, fax 603/891-0597, e-mail [email protected], or check URL http://www.wafernews.com.