TSMC commits $100 billion more to US chipmaking as AI demand surges

Taiwan Semiconductor Manufacturing Co. pledged an additional $100 billion to expand its US chipmaking capacity, bringing total American investment to $265 billion and signaling that the artificial intelligence infrastructure build-out is far from peaking.

"AI demand is not a one-year phenomenon — it is reshaping the semiconductor industry's capacity planning for the next decade," C.C. Wei, chief executive officer of TSMC, said in a statement.

The latest commitment comes as TSMC reported record quarterly profits, driven by surging orders from Nvidia Corp. and Advanced Micro Devices Inc. for chips used in AI training and inference. The company also raised its full-year revenue forecast, citing stronger-than-expected demand across its 3nm and 5nm process nodes (the manufacturing technologies that pack more transistors per square millimeter, improving performance and power efficiency).

For investors, the scale of TSMC's bet — $265 billion in US commitments alone — provides the strongest signal yet that hyperscalers and chip designers expect AI compute demand to compound for years. The question is whether the market has fully priced in the duration of this cycle.

What the $265 Billion Buys

TSMC's US expansion centers on three fabrication plants in Arizona, with the first facility already producing chips on the 4nm node. The second and third fabs are expected to manufacture 3nm and 2nm chips, respectively, with production ramping through 2028. The 2nm node (which delivers roughly 15 percent faster speeds and 30 percent better power efficiency than 3nm, per TSMC's published specifications) will be critical for next-generation AI accelerators from Nvidia, AMD, and Apple Inc.

The company's total US outlay of $265 billion dwarfs the roughly $45 billion it has invested in its home island of Taiwan over the same period, underscoring the geopolitical pressure to diversify manufacturing away from the Taiwan Strait. TSMC's main rival, Samsung Foundry, has committed about $20 billion to US chipmaking, while Intel Corp.'s foundry business has attracted roughly $10 billion in US government grants under the CHIPS Act.

Who Wins, Who Loses

The clearest beneficiaries of TSMC's capacity expansion are its largest customers. Nvidia, which relies on TSMC for its H100 and next-generation Blackwell GPUs, gains supply certainty for the AI chips that generated $47.5 billion in data center revenue last fiscal year. AMD, Broadcom Inc., and Qualcomm Inc. also depend on TSMC's advanced nodes for their AI and mobile processors.

For Intel, the news is more complicated. The company's foundry ambitions — built around its 18A process node — face an entrenched competitor with deeper pockets and proven manufacturing scale. Intel has said it expects foundry revenue to reach $15 billion by 2030, but TSMC's $265 billion US commitment suggests the Taiwanese giant is prepared to defend its market share aggressively.

Equipment makers stand to benefit as well. ASML Holding NV, the Dutch supplier of extreme ultraviolet lithography machines (each costing roughly $400 million), will need to deliver more systems to support TSMC's expanded fab capacity. Applied Materials Inc. and Lam Research Corp., which supply deposition and etching tools, also see a multiyear demand tailwind.

The Investment Case

TSMC shares trade at about 20 times forward earnings, a discount to Nvidia's 35 times multiple but a premium to the broader semiconductor index. The valuation reflects TSMC's unique position as the sole manufacturer capable of producing the world's most advanced chips at scale — a moat that the $100 billion commitment only reinforces.

The risk is execution. Building and equipping advanced fabs takes three to five years, and TSMC has already faced delays at its Arizona site due to labor shortages and permitting issues. If the US expansion runs behind schedule, customers may be forced to rely on Taiwan-based production for longer, reintroducing the geopolitical risk the investment was meant to mitigate.

Still, for investors looking for a direct bet on the AI infrastructure cycle, TSMC offers something no other company can: the physical capacity to build the chips that power the largest computing build-out in history. The $100 billion commitment is not just a number — it is a timeline.

This article is for informational purposes only and does not constitute investment advice.