TSMC's first Arizona fab began commercial production of 4-nanometer chips in early 2026 — a milestone that was simultaneously celebrated as a triumph of American industrial policy and dismissed as a drop in the bucket. The Arizona facility produces approximately 24,000 wafers per month, roughly 2% of TSMC's total global capacity and less than 5% of the advanced-node chips that Apple, Nvidia, and AMD consume annually.
The scale of the challenge is difficult to overstate. Taiwan produces more than 90% of the world's most advanced semiconductors — the chips below 7 nanometers that power everything from iPhones to F-35 fighter jets. A disruption in the Taiwan Strait, which military analysts consider a non-trivial risk within the next decade, would cause an estimated $1 trillion in global economic damage within the first year alone, according to a Boston Consulting Group study commissioned by the Semiconductor Industry Association.
The CHIPS Act's $52 billion in subsidies has accelerated the reshoring timeline, but the numbers reveal sobering math. Building a single advanced-node fab costs $20-30 billion and takes three to five years. To achieve true supply chain resilience — defined as the ability to meet global demand for advanced chips without Taiwan — would require at least eight to ten new fabs outside Taiwan, at a total cost of roughly $250 billion. The current pipeline of committed projects — TSMC's Arizona fabs, Samsung's Texas expansion, Intel's Ohio complex — covers only about a third of that requirement.
Advanced packaging, the process of integrating multiple chiplets into a single system, adds another layer of vulnerability. Nearly all of the world's advanced packaging capacity — including TSMC's CoWoS technology, critical for Nvidia's AI chips — is concentrated in Taiwan. Intel and Samsung are building packaging facilities in the U.S., but they won't reach meaningful capacity until 2028 at the earliest.
The geopolitical calculus is forcing difficult decisions in boardrooms across the technology industry. Nvidia has publicly committed to using TSMC's Arizona fab for a portion of its next-generation GPU production. Apple, which has been TSMC's largest customer for a decade, is reportedly negotiating a $12 billion prepayment agreement to secure Arizona capacity for its 2028 iPhone processor. But these commitments, while symbolically important, still represent single-digit percentages of each company's total chip consumption.
China, for its part, is investing over $150 billion in domestic semiconductor capacity through its "Big Fund" initiative, with a focus on mature-node chips (28 nanometers and above) where it can compete without access to advanced lithography equipment restricted by U.S. export controls. This strategy — dominate the less glamorous but enormous market for automotive, industrial, and IoT chips — could give China significant leverage in supply chains even if it never catches up at the cutting edge.