Semiconductors: Straight Talk from the Former Head of TSMC

Claude Barfield

Dr. Morris Chang is founder and former chairman/CEO of Taiwan Semiconductor Manufacturing Company (TSMC), the world’s largest and most advanced semiconductor manufacturing firm. Recently, he has caused great heartburn among the proponents (including within his own company) of the US spending tens of billions of dollars to “reshore” semiconductor manufacturing. Chang stated bluntly: “I think it will be a very expensive exercise in futility.” His statement comes as TSMC’s current leaders have announced plans to build a $12 billion semiconductor manufacturing plant in Arizona, and are maneuvering to compete for subsidies included in the $52 billion CHIPS for America Act moving through Congress as part of a larger technology and trade bill.

Born in China, Chang came to the US, studied at Harvard University and Massachusetts Institute of Technology, and obtained a PhD from Stanford University. He spent over 20 years at Texas Instruments (TI), ultimately heading the company’s integrated circuit (IC) programs, then was recruited to lead Taiwan’s Industrial Technology Research Institute, which was tasked with fostering cutting-edge technologies.

At the institute, Chang inherited a project for advancing ICs that included hundreds of engineers. He soon made plans to take the project out of the institute and establish a private company; while grateful for a government that supported technology, he believed the institute’s semiconductor team would always have a limited budget and, more importantly, not face serious competitive pressures.

Chang’s deep knowledge of the industry’s history and complexity paved the way for him to upend its mode of operation. Acutely aware that Taiwan could not compete on advanced chip design (“nodes,” in industry jargon), he built a business model based solely on manufacturing chips designed by other companies (“pure-play” foundries). “We did not compete with Intel, we did not compete with TI. We became their suppliers,” Chang explained. Chang and TSMC thus took advantage of a longer-term shift: the desire of many advanced IC companies to leave behind the tedious, expensive process of manufacturing. “If a foundry was willing to build mature nodes . . . [TI and Motorola] would be very happy to let the foundry do [this],” Chang continued.

From his own experience, Chang has developed strong faith that the efficiency of markets can foster particular competitive advantages and disadvantages. He argues that in the US and other advanced economies, engineering talent in the 1980s became increasingly attracted to design and other elements of the complex semiconductor industry—or to Wall Street—while in Taiwan, top-flight engineers flocked to TSMC and other advanced manufacturing companies. Over the past three decades, this manufacturing superiority has only increased and is the principle reason Chang is skeptical of the current US attempt to build semiconductor supply chains outside of Asia—and Taiwan. Citing TSMC’s experience with an Oregon plant, he asserts there is a 50 percent cost differential with semiconductor plants in Taiwan. But because TSMC is so advanced, it can still make some profit in Oregon.

From this history and a broader view of manufacturing competitiveness, he holds that the current US goals for domestic semiconductor manufacturing will be much more costly than projected. He also challenges the notion that a loss of manufacturing capability means the US semiconductor industry is in dire straits (key to the CHIPS Act debate). Chang points out that the US is doing quite well in other areas of the IC supply chain: “Right now, I think the United States has a very good position in semiconductor technology. . . . The US has got most of the design capability in the world, and the best design capability in the world.” As one of the Brookings hosts noted, his comments clearly contradict the concept of industrial policy that seems to be winning out in the US.

While eschewing political issues and avoiding criticism of the current TSMC leadership, Chang notes that TSMC’s decision to construct an Arizona plant came after his retirement and was made at the US government’s direct request. But on the Taiwan security issue, he was less than convincing. As to whether TSMC’s plant locations in Taiwan were unsafe, he posited that any conflict in the Taiwan Strait would generate “a lot more to worry about than just chips.” This is true, but Chang ignores the real possibility that short of conflict, China could cripple TSMC’s ability to manufacture the most advanced chips—of which the company is the world’s sole source. In sum, there is a compelling national security rationale for the US to build domestic capacity to manufacture advanced and mid-level chips. But Chang makes cogent points regarding the future economic viability of such plants. US leaders should be aware that the current rush for reshoring could well overshoot and produce “high-unit cost” plants that will be “non-competitive in world markets.”

Courtesy: (AEI.org)