Chipmaking exec: China to spark new bust
A rise in capacity triggered by China's arrival as a semiconductor manufacturer is likely to cause another chipmaking slump in 2005, according to the head of TSMC.
- Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Morris Chang, founder and chief executive of Taiwan Semiconductor Manufacturing Co., said Monday there have been busts in the chipmaking industry every five years, often caused by the entry of new countries into the market.
"Will there be a 2005 recession? Yes, I think there will be. And who will cause it? I think China," Chang told businessmen at the Taiwan and China Semiconductor Industry Outlook conference here. "China is building a lot of chip capacity."
Chang, one of the most powerful individuals in the microprocessor industry, had other gloomy views for chip partisans: Moore's Law is technologically, but not economically, sustainable, he said, because it's getting too expensive to build new chip fabrication facilities, or "fabs." The law, outlined by Intel cofounder Gordon Moore in the 1970s, posits that the amount of circuitry on a chip will double every 18 to 24 months, resulting in rapid decreases in the cost of delivering a given amount of computing power.
"Moore's Law will have to slow down," Chang said. "Economically, we're running into a wall here. We can't afford to do these expensive designs anymore. We can't afford a big fab anymore."
Fabs get more expensive as Moore's Law progresses, said to Chang, who said it takes $30 million to design a complex chip in the current generation of technology, referring to chips with features about 90 nanometers (billionths of a meter) in length.
"To outfit a fab to make these 90-nanometer chips, one needs to spend $3 billion," he said. "Very few companies can afford $3 billion to build a fab."
Analysts have become more optimistic lately about near-term prospects for the chipmaking business, which went through a severe slump that began in 2000. On Monday, Merrill Lynch analysts lifted their forecast for industry revenue in 2003 from 10 percent to 12 percent, predicting total sales of $157 billion.
For 2004, the analysts raised the growth estimate even more, increasing it from 15 percent to 18 percent, meaning total sales of $185 billion. In addition, Merrill Lynch gave a preliminary growth estimate of 21 percent for 2005--a $225 billion market.
A slump may be unavoidable, but it won't be as severe as some have been, including the "catastrophe" of 2000, Chang predicted. "It might be comparable to the 1990 one that was relatively mild," he said.
Moore's Law will slow to about half its current pace for the next 10 years, Chang predicted. After that, there still will be about 15 years or so of a "post-Moore's Law period," during which computer designers will finally be able to catch up with the pace of processor innovation, he said.
"I still think you have a quarter-century of technological innovation," Chang said.
Riding the boom-bust wave
A bust begins when overcapacity causes prices to drop, not when demand for chips dries up, Chang said. Chipmakers that suffer typically wait about two years before investing in new capacity. After that--because product orders recover and keep growing--the next phase of construction begins again.
New busts have been spurred in the past by the arrival of Japan, Taiwan and Korea in the semiconductor manufacturing industry, Chang said. China will be the next trigger, he added, saying he suspects the new slump will arrive in late 2005, though it might be in early 2006.
For a chipmaker, dealing with a bust is a no-win situation, Chang said.
"When you know as a businessman there is going to be capacity glut, but when your competitor is recklessly adding capacity, what do you do?" he said. "I have been wrestling with that for 40 years."