The global semiconductor industry is undergoing significant shifts, primarily driven by geopolitical dynamics, including the U.S.-China rivalry, and increased concerns over supply chain resilience. Taiwan Semiconductor Manufacturing Company (TSMC), a leading producer of advanced microchips, is expanding its manufacturing operations outside Taiwan. Simultaneously, the United States is actively incentivizing domestic chip production, while China intensifies its pursuit of technological self-sufficiency in its domestic chip sector. These developments underscore the strategic importance of semiconductors, which are critical for military technologies and artificial intelligence, and coincide with intelligence assessments regarding the security of Taiwan.
TSMC's Global Expansion and Strategic Rationale
Taiwan Semiconductor Manufacturing Company (TSMC), headquartered in Hsinchu, Taiwan, has historically produced over 90% of the world's most advanced microchips. The company has initiated a strategic expansion of its manufacturing operations beyond Taiwan, citing customer demand, a focus on technology leadership, manufacturing excellence, and customer trust as primary drivers, according to Chief Financial Officer Wendell Huang.
Arizona and Beyond
In 2020, TSMC announced plans for semiconductor fabrication plants (fabs) in Arizona, U.S. The first Arizona fab commenced high-volume production in late 2023. Plans for Arizona include a total of six fabs, two advanced packaging facilities, and a research and development center. TSMC CFO Huang noted that 70% of the company's revenue originates from the U.S. and that U.S. customers require advanced technology, necessitating the expansion of advanced technology fabs in the region. The Biden administration's CHIPS Act provides incentives for increasing chip production within the United States. Beyond the U.S., TSMC is also expanding its presence in Japan and Germany.
TSMC operates on a "pure play foundry" model, manufacturing chips designed by client companies such as Apple, Sony, and Nvidia. Huang stated that overseas expansion also allows access to broader talent pools, land resources, water, and power, addressing potential resource limitations within Taiwan.
Despite this global expansion, TSMC intends to maintain its primary base and advanced research facilities in Taiwan, with its most advanced processes expected to remain on the island for the foreseeable future.
The company reported a revenue increase exceeding 30% and a profit increase approaching 40% year-over-year in its latest quarterly report, with high-performance computing, particularly for artificial intelligence applications, identified as a significant growth driver.
U.S. Initiatives for Domestic Chip Production and Supply Chain Security
The United States has identified its reliance on Taiwan for semiconductors as a significant vulnerability, according to National Security Adviser Jake Sullivan.
In July 2023, a classified CIA briefing for prominent tech executives reportedly conveyed intelligence suggesting a potential Chinese attack on Taiwan by 2027. This briefing was reportedly arranged to encourage the tech industry to diversify chip production away from Taiwan.
CHIPS Act and Investment
The U.S. government has aimed to encourage domestic semiconductor manufacturing, leading to the enactment of the CHIPS and Science Act of 2022, which allocates $50 billion in government subsidies for building U.S. plants. While domestically produced chips have been reported to cost over 25% more than those from Taiwan due to higher material, labor, and permitting costs, industry commitments have been made.
Apple CEO Tim Cook has since committed to investing $100 billion in the United States to support TSMC and other chip manufacturers. TSMC itself has pledged approximately $165 billion in U.S. investment, including land for at least five additional plants in Phoenix. Nvidia's first U.S.-made AI chip was produced at TSMC's Arizona facility, though advanced packaging for these chips currently requires shipment to Taiwan. TSMC's Arizona plants currently utilize technology a generation behind what is available in Taiwan.
Economic Risks and the "Silicon Shield"
Confidential reports have highlighted potential economic consequences of a disruption to Taiwan's chip supply.
A 2022 report indicated that losing access to Taiwan's chip supply could trigger an economic crisis, potentially causing an 11% drop in U.S. GDP. A January 2024 Bloomberg report estimated a conflict could cost the global economy over $10 trillion.
Taiwan's government maintains an unofficial policy, known as the "silicon shield," which requires TSMC to keep its most advanced manufacturing technology on the island, aiming to make Taiwan economically vital to deter attack. However, some analysts note that events such as Russia's invasion of Ukraine have demonstrated that economic self-interest does not always prevent military aggression.
China's Pursuit of Semiconductor Self-Sufficiency
The United States has implemented export controls aimed at limiting China's access to high-end microchips and chip manufacturing equipment. These restrictions are intended to prevent China from achieving superiority in artificial intelligence and to restrict the Chinese military's access to advanced chips.
In response to these restrictions, China has intensified its focus on technological self-sufficiency within its chip industry. The Chinese government has allocated over $200 billion towards developing a modern and independent domestic chip sector, with companies like SiCarrier, specializing in epitaxy equipment and atomic layer deposition tools, identified as key players. Chinese leader Xi Jinping stated in April that scientific breakthroughs are necessary for China to achieve superiority in AI and pledged substantial policy support for these endeavors.
Domestic Challenges and Lag
China's domestic chip industry currently lags behind global leaders, particularly in high-end AI chips and capabilities for mass production of advanced semiconductors. Challenges identified by industry experts include:
- Restrictions on acquiring advanced graphics processing unit (GPU) chips for AI data centers.
- Limitations on purchasing high-end chipmaking equipment.
- Constraints related to cloud computing access.
- AI software predominantly optimized for foreign chip architectures.
These challenges necessitate innovative development pathways and significant time and resources for progress, as indicated by Jarod Wang from a Shenzhen-based chip design firm.
Recently, the U.S. administration adjusted some chip restrictions, permitting California-based Nvidia to sell its H200 GPU, an advanced but not top-tier chip for AI data centers, to select buyers in China. U.S. officials have suggested that providing access to these chips could maintain China's reliance on U.S. technology and potentially reduce its drive for complete chip independence.
However, experts like Chris Miller, author of "Chip War," express skepticism regarding this approach, noting China's consistent pursuit of its own chip ecosystem for over a decade. David Sacks, a China specialist at the Council on Foreign Relations, highlighted a potential strategic risk in supplying advanced AI tools to a principal rival, considering AI's transformative impact on economies and military capabilities.