TSMC CEO CC Wei was taking questions at an event when Nicky Lu, vice chairman of Monte Jade Science and Technology Association, recalled that a few years ago, TSMC founder Morris Chang led a delegation of key executives of the company to a global semiconductor conference in South Korea. Chang quipped that Samsung may be a giant, but as along as you can find the chink in its armor, you can still hurt it.
My question for Wei, who was making a speech at the event organized by Monte Jade Science and Technology Association, was that while TSMC’s capex over the past few years has stayed at about 45% of its revenues, Samsung’s foundry service sales have been about 30% of TSMC’s total sales. If that’s the case, Samsung would have to raise its capex to 150% of its sales in order to compete with TSMC. But would that be feasible?
I further asked if TSMC would adjust its capex strategy as the global foundry sector might see a decline in 2023. Chang once noted that he was a “learning curve believer,” meaning that he would keep the pace of investment when he was ahead, leaving competitors far behind. But under what circumstances will TSMC adjust its capex strategy in the future?
Wei said that TSMC’s capex is based on a two-and-a-half-year cycle, and that the current investment is meant to cater to customers two-and-a-half years from now. Is that true? Then why did TSMC reduce its 2022 capex to US$36 billion from the original plan of US$42 billion? I think the key is that when short-term orders shrink, capex will also be cut accordingly. But Wei wouldn’t disclose the reasons behind the capex reduction.
Global demand for semiconductors will continue to grow, and the cars of the future will just be a mobile supercomputer on four wheels. Business opportunities are everywhere, from artificial intelligence (AI) to 5G and others. But the world is moving from globalization to regionalization, and we expect safer, greener, and better ecosystems.
But these expectations will also bring higher costs. Wit the rules of the game changing, will Samsung still be able to compete with TSMC? I don’t think so. Instead, Intel, with policy support from the US government, will have a fatal impact on Taiwan’s advantages.
TSMC’s fab project in Japan is meant to satisfy the needs of big customers there, but it is also important for TSMC to recruit talent from Japan. When Japan explored the possibilities concerning TSMC’s local fab project, a complete technological transfer package was among the options. But the Japanese side did not think they would have the sufficient workforce to fully meet TSMC’s demand, though recruitment of some manpower from Japan could still be possible.
You may have heard that TSMC has sent more than 500 people to Arizona to work at its US fab, but you probably are not aware that over 200 people from Japan are being trained at a TSMC facility in Taichung. They will be the core staff of the Japan fab. Taiwan and Japan will see coopetition between them, but I think there will be more on the cooperation side.
For some of the questions, Wei wasn’t really answering. But this is what the relations between the media and businesses are supposed to be. A high-profile businessman has its corporate and social responsibilities to fulfill, and there are protocols and rules to be followed when it comes to disclosing a company’s strategies. What the media should be doing is to make preparations before covering an event, and find the clues from what the speaker says.
From Wei’s speech, I got a clearer picture of TSMC’s investment strategy in Japan, its views on US-Japan cooperation, and its opinions about regional division of labor. As for the other comments, I don’t have much to say.