Issue



Big surge coming forecasts TSMCs Morris Chang


12/01/1998







Big surge coming, forecasts TSMC`s

Morris Chang

Al Furst, Asia Consultant, PennWell International

As semiconductor slumps go, this one is relatively mild, and it will be followed by a big surge in business, predicts Morris Chang, chairman of foundry-leader Taiwan Semiconductor Manufacturing Co. (TSMC), and a veteran of several industry downturns.

In each of the past four to five years, PC growth has exceeded 15%, and even now unit sales have probably slowed only to the 12% range, Chang pointed out in an interview at TSMC`s headquarters in Hsinchu, Taiwan. Dataquest reports 3Q98 worldwide PC shipments were up 13.7% from a year earlier. What compounds the current slump, however, is the rapid drop in average selling price, lowering dollar sales, and thus eating into chip margins in the largest market for semiconductors. The decline was also aggravated by an inventory overhang when PC sales slumped in the fourth quarter last year, which he believes took about six months, and maybe more, to work down. He sees this as "a short-term temporary saturation of the PC market."

Similar slowdowns occurred in the early `70s, the early `80s, and again starting in about 1992. Each boom that followed was fueled by special applications. The boom back in 1972-75, for example, was triggered by the emergence of electronic games, calculators, and digital watches. When those products began to saturate the market, there was a bust even more serious than we are experiencing now. The boom in 1982-85 marked a surge in sales of personal computers, following IBM`s entry into the PC market in 1981. Many new PCs and even new PC companies emerged at that time, leading to the bust of 1985-86, which Chang also feels was worse than the current downturn. The last boom cycle, which lasted until about 1992, he says, was marked by tremendous growth of the use of PCs in businesses as well as the home.

High-risk capital management. To put the current situation in perspective, however, Chang says it is necessary to look at not just slowing demand, but the supply side as well.

"The supply side problem becomes bigger and bigger because the capital intensity in the semiconductor market has progressed rather rapidly," he points out. This is not a problem for companies that manage capital well, but it becomes a serious issue for those that haven`t had experience in capital management.

"In Korea, they have been following a high risk, what I would consider a reckless path, for quite a few years," Chang says. "They are highly leveraged. In the chaebols, the ones that count, debt/equity ratios are about 9:1- 90% debt, 10% equity."

As less money comes in, they need to expand credit even further, but as capacity slumps, "they have a hard time paying interest on the debt they have already incurred. That`s what highly leveraged companies face when there`s a downturn." The banks won`t loan them any more money unless the government steps in, and in Korea, that is not possible now.

"That`s the big difference between Taiwan companies, in general, and the risky approach Korean companies followed," Chang says. "We have always adopted a very conservative financial structure." In TSMC, in particular, he says, capital is about 35% debt and 65% equity.

"We have been expanding by using the money we made, so when a slowdown came, we cut back our spending because we had more capacity than we needed."

TSMC`s capital budget this year has been trimmed about 30% from earlier plans, he said, from US$1.3 billion to $930 million.

"On the operating side, you have to become efficient, a low-cost producer - and that we are," he says. When there is a slow-down, every company`s volume drops, but if you`re the low-cost producer at any volume, you are less hurt than other companies, and you can remain competitive, he explains.

What comes next? "You mentioned that past applications have caused the semiconductor industry to emerge from downturns, the consumer electronics of the `70s, and the PCs of the `80s. What will be the next engine of growth?" Chang was asked.

While PCs remain the biggest market, "the progression of the use of semiconductors in communications is steady and rising," Chang says. "I think there will be a big jump - something akin to the `92-`95 situation - when multimedia and the internet and the applications around them surge."

Although technical journals and trade magazines are discussing these applications right now, he said, they are still a limited phenomena and have not yet penetrated to the masses. These new applications also have not yet been adopted by big US businesses with large networks of PCs, let alone by medium-and smaller-sized companies. When this happens, Chang believes the market will surge as businesses interlink large numbers of PCs and network PCs.

Meanwhile, the slowdown has set back the adoption of 300-mm wafers for a year or even longer. To remain competitive, TSMC has to invest heavily in R&D. "Not product R&D," Chang explains. "We do not compete with our customers."

Instead, the company invests only in process R&D, and TSMC will spend close to US$100 million this year, he says. This can allow a foundry like TSMC to become a benchmark in the world semiconductor industry for process technology, he says. By contrast, semiconductor companies might spend 66-75% of their R&D budgets on product design.

Chartered implements CMP

Chartered Semiconductor`s CMP program is seeing rapid growth as it brings up 0.35-?m and below processes, but the Singapore foundry is finding that a lack of local consumables support has become a critical issue in implementing the CMP process throughout its fabs. Jiazhen Zheng, Chartered section manager for R&D/CMP, cited several support issues of concern, including long response and repair times, lack of local warehouse supply, and insufficiently trained field service engineers and support engineers. Moreover, cost of ownership rises with the lack of local supply.

In another development, Chartered and Pivotal Technologies LLC, Pasadena, CA, have signed a joint marketing agreement that allows the two companies to comarket analog and mixed-signal IP developed by Pivotal for Chartered`s foundry processes. Pivotal will offer its OptiMix line of silicon-ready analog and mixed-signal components, such as phase-lock loops, crystal oscillators, charge pumps, and other analog interface support elements for system-on-a-chip designs.- P.N.D.

NEC plans DRAMs for Shanghai fab

NEC is preparing to install a complete set of DRAM wafer processing equipment at its LSI plant in Shanghai, China, which is jointly owned with a Chinese electronics firm. The 216 ? 97 m main building has a total floor space of 77,000 m2. Work has proceeded to include 12,650 m2 of cleanrooms inside the building. NEC has purchased an additional 100,000 m2 of land adjacent to the existing plant for future expansion.

The new fab will be the most advanced line in China and this initial production with 0.35-?m devices is seen as an exercise designed to ease the transition to 0.25 ?m in the near future. Plans call for manufacturing standard 64-Mbit DRAMs with 0.35-?m design rules to start as early as February. Initial production capacity will be about 5000 wafers/month. The initial workforce will be 500 people including 40 Japanese from NEC in Japan. Manufacturing of logic LSIs and ASICs is slated to start at the end of 1999. The tentative final capacity is 20,000 wafers/month with 800 people. - P.N.D.

Komatsu reduces capital spending

Komatsu Ltd. is expected to reduce capital spending on its silicon wafer, polycrystalline silicon material, and excimer laser electronic business units this year, and is exploring the possibility of shedding some of its "idle assets," according to a rating report from Standard & Poor`s CreditWire. With chip markets still in a downward trend and demand for construction machinery softening, S&P said the rating outlook for Komatsu remains negative; the agency has affirmed its long-term A- rating for the company.

In FY97, Komatsu`s electronics business accounted for 9% of total sales. While a diversified revenue base has allowed the company to improve profitability in recent years, FY97 capital spending - primarily for the electronics business - exceeded funds from operations and is expected to do so again in FY98.