,) Taiwan's second-largest semiconductor company. IBM (IBM
) recently announced that it's losing money in its foray into the foundry business, which traditionally has been dominated by UMC and its bigger rival, Taiwan Semiconductor Manufacturing Co. (TSMC
). (Foundries manufacture semiconductors for other customers.)
Not only that, but the head of UMC's biggest customer, Xilinx, now questions whether IBM can compete with UMC. "IBM is strong in technology, but not in manufacturing," says Willem P. Roelandts, the president and chief executive officer of Xilinx (XLNX
), which uses UMC and IBM to make its advanced chips used in consumer electronics and telecom and networking equipment. UMC is "more disciplined in improving yields quickly" he adds.
UMC makes chips faster, too. If Xilinx needs new production urgently, the Taiwanese can deliver in about a month. "IBM is 50% to 100% slower," says Roelandts. The bottom line: UMC will remain his company's top choice.
HIGH SPIRITS. The foundry business is a tough one, requiring billions of dollars in facilities and years of experience in technology. Other than UMC and TSMC, the industry has few successes. Even the Singapore government, which has ample resources and a strong determination to be a tech leader, hasn't been able to turn Singapore's Chartered Semiconductor Manufacturing Ltd. into anything more than an also-ran.
Initially, it looked like IBM's venture would be a different story. And UMC seemed to be particularly vulnerable, since it's much smaller than TSMC, which gets far more -- a total of 17% -- of its revenue from the lucrative production of high-end chips with circuit widths of 130 nanometers. UMC, in comparison, gets just 6% from that same niche. And it's that high end of the market where IBM was aggressively staking its claim with Xilinx, the Silicon Valley chip designer that now remains UMC's biggest customer.
Small wonder Hu was in such high spirits when he spoke to me recently from his office in Hsinchu, Taiwan's Silicon Valley. He noted that the room smelled of orchids -- sprinkled among the many flower arrangements he received from well-wishers to congratulate him after taking the belm last month. "The florist business is doing quite well," Hu quipped. Now he can hope the same for UMC's fortunes.
"NEW CHALLENGES." Hu, 54, is a newcomer to UMC. Born in Taiwan to parents who had fled with Chiang Kai-shek's troops following the communist takeover of China, he studied electronic engineering as an undergraduate in Taipei before leaving for the U.S., where he got his PhD at the University of Illinois. Then he worked in Silicon Valley for 25 years before returning to Taiwan in late 2002.
As a relative newcomer to the foundry business, Hu is betting that both his outside experience and his background in design will prove pivotal as the chipmaking industry changes. Semiconductors are becoming ever more complicated, with circuits becoming narrower and narrower. All this "generates many new challenges and difficulties," Hu explains, which require the foundries to work more closely with their customers, the chip design companies.
Skeptics abound. Many investors have found UMC a perplexing company, which seems to be shifting to a new business model by investing in more chip design companies. Some people have wondered whether the foundry might try to concentrate on just a few key customers at the expense of others.
EVERYBODY GROWS. Is UMC departing from the foundry model? "We never said that," Hu says. "We are an integrated-circuit foundry, period." He says UMC will be selling off stakes in noncore investments such as a Taiwanese producer of white goods and a local financial-services company. "We're going to divest in those areas," he promises. As for the stakes that UMC has in chip design companies, "we will gradually sell those shares," he says.
Still, UMC isn't going to abandon its investment strategy altogether. Hu says he'll try to forge winning partnerships with other chip designers. He and his boss, UMC Chairman Robert Tsao, believe they must pursue partners if UMC is to have a chance against the likes of IBM. And at the commodity-level, low end of the business, new foundries in China and Malaysia are popping up to compete with UMC.
"We'll still invest in good design companies to make sure that we have good potential customers in certain technology areas," Hu says. "We have to pay special attention to startup companies that are technology leaders, so when they grow, we grow."
Given the volatile nature of the chip industry, Hu has his work cut out for him. But with IBM's recent setback in the foundry business, he has a little less reason to worry. Einhorn covers technology from Hong Kong for BusinessWeek. Follow his weekly Online Asia column, only on BusinessWeek Online