By Bruce Einhorn Morris Chang is the chairman of Taiwan Semiconductor Manufacturing Co., the world's leading made-to-order producer of semiconductors. He recently spoke with Bruce Einhorn from TSMC's headquarters in Hsinchu, Taiwan, about the global tech recession and how TSMC plans to emerge from the slowdown. (For more on TSMC's strategy, see BW, 4/30/01, "Betting Big on Chips"). Here are edited excerpts of that conversation:
Q: Last year, TSMC was operating at over 100% of its capacity. Today, you're around 70%. How bad will it get?
A: I still think that the third quarter will be better than the second and that the fourth quarter will be better than the third. My feeling is that it will be a more gentle recovery than what we experienced in 1998. Back then, in the first half, it got to as low as some 50% utilization. This time things are worse. I have a feeling that the bottom will be a little lower than last time, but we don't know yet.
Q: One reason you have so much unused capacity now is that TSMC spent $3.8 billion last year expanding capacity. Any regrets about that spending spree?
A: We could have slowed down a little last year. In retrospect, we spent too much. If we had spent $1 billion less, we would be in better shape. But my goodness, hindsight is 20/20. At the time, not a single voice was telling us that demand wasn't going to be there this year.
Q: Some people are expressing doubts about the foundry industry. What's your response?
A: The long-term trends remain unchanged. The future is still very great.
Q: You have a joint-venture fabrication facility in Singapore, and your big rival, United Microelectronics Corp., is building a fab there, too. Do you think Singapore is a good choice for Taiwanese looking to branch out beyond Taiwan?
A: We don't see Singapore as a place where we will have a large cluster of fabs. It has a population of only 3 million. And the human resources are even poorer than the size of the population would indicate. The technology resources in Taiwan are considerably greater.
Q: How about the U.S.? TSMC has a subsidiary in Washington State called WaferTech.
A: I see a possibility for building a large cluster of plants [in the U.S.]. The latest recession has set back that plan -- we don't need any more fabs outside of Taiwan right now. But still I think that WaferTech will be the nucleus of more fabs in the U.S. in the next five years.
Q: The Taiwan government forbids you to invest in China. But if the government drops that ban, are you interested in the mainland?
A: Mainland China is certainly a very strong possibility. We are not going to China in the next three to five years, but in this business you have to plan ahead. It takes a year just to find a satisfactory site. It takes two years to build up a factory. After that, it takes two years to ramp up. So it's a five-year proposition for a new site. [T]he government...ought to let us go in three to five years.
Q: So if the government relents, you're ready to move to China?
A: China is not the only place. It's a contender. In China, you read a lot about political changes. But there is still a lot of uncertainty -- corruption and so on. If it continues to get worse, I think that would make [the mainland] less attractive. Also, there is a workers' culture problem. Turnover there is pretty humongous.
Q: Other Taiwanese have taken the plunge, investing through non-Taiwan companies. What do you think of those plans?
A: I have my doubts. Richard Chang [of Semiconductor Manufacturing International Corp.] is building a community complex -- not only [worker] dorms but also a school and a country club. I don't think that's all good, the paternalistic thing. I thought they were moving away from that. That seems to be moving back to the old way.
Q: Overall, what's your view on the new companies entering the business?
A: There is too much money around. I'm not too concerned about them yet. I would be concerned if they became bigger, but now they are just one fab each. I really don't want to allow them to grow. We have to prevent that from happening.
Q: More specifically, what's your opinion about the two new foundries that have opened in Malaysia?
A: They don't seem to be going anywhere, frankly.
Q: How about the Koreans? Since the market for memory chips is so bad, Hyundai has said that it is expanding its foundry business.
A: I'm less worried about the Koreans. Though their size is big, their culture is very different. They are now 100% in the commodity culture [of DRAM chips]. In the commodity culture, the important thing is cost. Service is not important at all. The Koreans are pretty far distant from us in terms of service, and without a service component, they can't be a serious threat to us.
Q: If demand continues to slump, does TSMC have any interest in getting into memory chipmaking?
A: No, not really. We do some DRAMs now. We are still very interested in the embedded DRAM area, because we think that it is part of the system-on-a-chip. In order to be a leader in system-on-a-chip, we need to know quite a bit about DRAM manufacturing. But in the stand-alone DRAM market -- the commodity DRAMs that you have -- we have limited our exposure to that area to less than 15% of our capacity.
Q: With the industry getting more crowded, how do you see TSMC keeping its edge?
A: My role model is the investment bank. An investment bank has a lot of competitors. To keep your loyalty, they jump through hoops for you.
Q: But some people say that good execution will soon become just another commodity. In other words, everybody will be able to offer quality service and technology. What do you do then?
A: The commodity thing is exactly the opposite of what we want. I think we realized the danger a few years ago that the foundry service might become a commodity. We have in the last few years trying to steer ourselves away from that.
A: Build relationships. For the past few years we have been trying to cultivate the partnership concept. Not every customer can be a partner, but a few can be. A commodity means that the buyer can switch loyalties overnight, the buyer can pick and choose whoever produces the cheapest and highest-quality product at the moment. If you have a partnership established, then it's not a commodity anymore.
I have been using investment banking as my role model. The same kinds of services are available from maybe a dozen investment banks. But you don't think of investment banking as a commodity mainly because of relationship and service. When they provide you with good service, and a good relationship, then you tend to stick with that bank.
I want to provide the same quality of service. We jump through hoops for our customers -- at least, for some of our best customers. That's the kind of thing that doesn't become commoditized. Einhorn covers technology from Hong Kong for BusinessWeek. Follow his weekly Online Asia column, only on BW Online