TSMC invests another $100 billion to expand production in the US: A milestone in the restructuring of the semiconductor industry chain

Event Overview

In February 2026, TSMC announced an additional $100 billion investment in Arizona, USA, to build four more wafer fabs. TSMC had previously invested approximately $40 billion in two fabs in Arizona, bringing the total US investment to $140 billion, making it the largest single overseas investment in the semiconductor industry. The new fabs are expected to produce 3nm, 2nm, and more advanced process chips, and may incorporate advanced packaging capabilities.

Background Analysis

• TSMC's expansion is driven by multiple factors:
• Geopolitical pressures: The US CHIPS and Science Act provides substantial subsidies, while tighter export controls force TSMC to disperse advanced capacity to the US.
• Customer demand: Major US customers like Apple, Nvidia, and AMD seek to enhance supply chain resilience and reduce dependence on a single fab in Taiwan.
• Technology competition: Intel Foundry and Samsung Foundry are aggressively catching up, and TSMC needs to solidify customer loyalty through capacity expansion.

In-depth Analysis

Technology Impact

• The advanced technology roadmaps involved include:
• N2 (2nm) process: TSMC plans to deploy its 2nm GAA (Gate-All-Around) process at US fabs, with technical barriers lying in multiple exposures of extreme ultraviolet lithography (EUV) and material innovations.
• 3D IC and advanced packaging: To meet AI chip interconnect demands, the fabs may integrate InFO (Integrated Fan-Out) and CoWoS (Chip-on-Wafer-on-Substrate) technologies, placing higher requirements on wafer-level packaging equipment and materials (e.g., temporary bonding adhesives).
• High-temperature, high-reliability processes: Some fabs may be used for military and aerospace chips, requiring special process certifications.

Technical barriers: TSMC has deep expertise in yield control and equipment calibration, but the US lacks skilled technicians, which may temporarily slow production ramp-up.

Supply Chain Impact

• Upstream equipment vendors benefit significantly:
• ASML: Each 5nm-class fab requires about 30-50 EUV lithography machines. TSMC's expansion will drive orders for its High-NA EUV, expected to boost ASML's revenue by 20% before 2027.
• Applied Materials, Lam Research, KLA: Demand for etching, deposition, and inspection equipment surges, with US equipment vendors gaining priority for local orders.
• Material suppliers: Wafer makers Shin-Etsu Chemical and SUMCO, photoresist companies JSR and Tokyo Ohka Kogyo, and specialty gas firms (e.g., Linde) will establish local delivery capabilities in the US.Changes in Midstream Foundry Competitive Landscape:
• If TSMC's capacity in the US is fully utilized, it will account for 10-15% of its total capacity, reducing the impact of a single earthquake risk in Taiwan.
• Intel Foundry Services (IFS) will face more intense competition, especially as major customers like Apple and AMD may prefer to cooperate with TSMC's US factory.
• Samsung Foundry already has a factory in Texas, USA, but its technology node lags behind and its investment scale is not as large as TSMC's.

• Downstream Packaging and Testing:
• Packaging and testing companies such as ASE and Amkor are expected to build supporting capacity in the US, forming a "wafer manufacturing + packaging and testing" cluster.

Competitive Landscape

• TSMC vs Intel: Intel plans to leverage its US identity to secure government orders, but TSMC still holds an advantage with its technological leadership and customer relationships. Intel needs to achieve a yield breakthrough on its 18A process.
• TSMC vs Samsung: Samsung has taken the lead in mass production of 3nm GAA, but yields are low and it lacks major customers. TSMC's expansion in the US will further compress Samsung's foundry market share.
• Customer Lock-in Effect: AI chip companies like NVIDIA and AMD are expected to obtain a "Made in USA" label, enhancing political and market compliance.

Regional Implications

• United States: The self-sufficiency rate of chip manufacturing will increase from currently less than 10% to about 20% by 2030, but costs are 30-50% higher than in Asia, requiring continuous subsidies.
• Taiwan, China: Faces concerns of "de-TSMC-ization," but the concentration in Taiwan cannot be shaken in the short term—90% of global advanced processes are still produced in Taiwan.
• China: The US strengthening of domestic manufacturing will intensify technology controls, making it more difficult for China to build advanced capacity on its own, which may accelerate expansion of mature processes and development of domestic EDA/SiC.
• South Korea and Japan: South Korea needs to consolidate its memory advantage, while Japan is catching up in advanced processes through Rapidus.

Investment Perspective

The capital market remains confident in TSMC's long-term growth, but faces cost pressures in the short term. Analysts expect that the US factory will increase TSMC's capital expenditure as a percentage of revenue from 30% to 40%, but customers are willing to pay a premium. Equipment stocks (ASML, Applied Materials) and materials stocks (Shin-Etsu Chemical) will continue to benefit.

Long-Term Outlook## Long-Term Outlook

Over the next 5-10 years, global semiconductor manufacturing will present a "bipolar" landscape: 1. Advanced processes: Concentrated in Taiwan, the United States, South Korea, and Europe (Intel/TSMC collaboration). 2. Mature processes: China and Southeast Asia will capture a larger share. TSMC's expansion in the U.S. may spur Samsung and Intel to accelerate investment, but the technology gap persists. In the long term, whether the U.S. can build a complete wafer fabrication ecosystem (including chemicals, gases, and talent) is the biggest uncertainty.

Conclusion

TSMC's $100 billion investment is not just a corporate strategy but a microcosm of the global semiconductor supply chain's shift. It will accelerate the dispersion of advanced processes, solidify TSMC's core position in the AI era, and force countries to reshape their domestic chip industries. For equipment and materials suppliers, this is a once-in-a-decade opportunity; for competitors, it is a challenge that must be answered.