The performance of Taiwan's TSMC, the world's largest foundry (semiconductor contract manufacturing) corporation, is starkly contrasting between its plants in the United States and Japan. The Arizona plant has successfully turned a profit in its first quarter of mass production by securing major clients like Apple, whereas the Kumamoto plant in Japan has recorded a deficit in the hundreds of billions of won due to insufficient demand, with an operational rate capped at half.
According to Taiwan's Commercial Times on the 18th, TSMC's Arizona plant recorded a net profit of 42.32 billion Taiwan dollars (approximately 195 billion won) in the second quarter, marking its second consecutive quarter of profit. This came just one quarter after starting mass production of 4nm (nanometer, one billionth of a meter) technology in the fourth quarter of last year. In contrast, TSMC's Japan JASM plant incurred a net loss of 29.73 billion Taiwan dollars (approximately 137 billion won) in the second quarter, with a cumulative deficit of 45.2 billion Taiwan dollars (approximately 208 billion won) for the first half of the year.
The fates of the two plants hinge on the presence or absence of "anchor customers". The Arizona plant was aimed at major U.S. tech clients like Apple from the planning stage. Supported by U.S. government policy to attract advanced semiconductor supply chains amid the U.S.-China technology hegemony war, TSMC is investing $165 billion (approximately 228 trillion won) in the Arizona project. Following Apple and AMD's firm bookings for a monthly supply of 30,000 4nm wafers, NVIDIA has also started securing some volume, helping the operational rate of the first plant reach 100% in a short period.
Industry evaluations suggest that TSMC's local production in Arizona has stabilized on a solid trajectory. Currently, construction of the Arizona plant's second phase is nearing completion, with plans to begin bringing in equipment in the third quarter of next year and initiate mass production of 3nm technology.
In contrast, the Kumamoto plant opted for a different strategy from the outset. When announcing investments in 2021, TSMC designated Japan's automotive and electronic component industries as key customers and brought in joint venture partners like Denso and industry leader Sony for automotive semiconductors and image sensors. The Japanese government, aiming to restore the glory of its past semiconductor kingdom, supported about 40% of the total investment, while TSMC sought to leverage Japan's global materials, parts, and equipment ecosystem. The Kumamoto plant, which began operations in February this year, focused on legacy (mature) processes ranging from 12 to 28nm.
However, the legacy processes are marked by intense competition and low profitability, and the volume from Sony and Denso alone has proven insufficient for stable plant operation. Failure to secure additional large customers led to an operational rate hovering at around 50% in the first half of the year, directly translating the low operational rate into operational losses. Industry insiders noted, "Subsidies only lower construction costs, while the profitability of the plant ultimately depends on the operational rate."
Plans to expand the second plant in Japan are also slowing down. Initially, TSMC planned for a 6nm process at the second plant, but there have been concerns that by the scheduled operational date in 2027, advanced processes below 2nm will already be in mass production in Taiwan, potentially diminishing its competitiveness. The sluggish recovery of the automotive and consumer electronics markets is another factor contributing to demand uncertainty for the Japan plant. An industry official remarked, "Japan lacks core customers needing advanced processes, making both demand and technological positioning ambiguous. The Kumamoto plant will struggle for the time being."