台積電位於美國亞利桑那州的晶圓廠，近期成為全球半導體產業討論的焦點。最新財報顯示，該廠的獲利情況出現劇烈下滑，從2025年第二季的42.32億新台幣暴跌至第三季僅剩4100萬新台幣，幾乎等同於獲利蒸發99%。這種斷崖式的下降，引發外界對台積電赴美建廠是否「得不償失」的強烈質疑，也讓其在美投資超過1.2兆新台幣的長期回本計畫蒙上陰影。

台積電選擇在亞利桑那建廠，原本是為了配合美國政府推動半導體製造業回流的政策，同時藉由在地化布局減少供應鏈風險。然而，理想的全球戰略在落地過程中卻遭遇層層挑戰，其中最關鍵的因素就是美國本地製造成本遠高於預期。從建廠工程延誤、成本超支，到技術人才難以就地補充、工會制度造成的管理成本上升，都讓這個海外投資案承受巨大壓力。

造成獲利大幅縮水的核心原因之一，是亞利桑那二廠正在加速向3奈米與其他高端製程轉型。與主攻成熟製程、運行相對平穩的一廠不同，二廠的任務是滿足在AI熱潮帶動下，全球對尖端晶片爆炸性成長的需求。然而，越先進的製程意味著需要更昂貴的設備、更複雜的工藝，以及更高水平的技術人力。這些投入不僅讓資本支出飆升，同時快速推高營運成本，使獲利能力遭到沈重侵蝕。

此外，在美國製造晶片的基本成本結構本身就十分不利。當地的建築費、機台搬運費用、保險成本等幾乎全面高於台灣，日本或南韓；技術人力更是必須部分依賴外派工程師支援，造成薪資與訓練費用同步增加。這些結構性問題不是短期內可以解決的，使台積電在美的投資在未來數年都可能面臨盈利不如預期的情況。

儘管美國市場對其長期策略具有重要意義，但從現階段的財務數據來看，台積電美國事業體的獲利能力仍然遠落後於台灣及其他海外基地。隨著3奈米與更先進製程的持續投入，短期內恐難見明顯改善。外界也開始關注，台積電是否需要重新調整其在美布局策略，以避免龐大投資成為長期財務包袱。

整體而言，亞利桑那工廠獲利暴跌的情況，已不再只是單一企業的經營問題，而是反映全球半導體產能在不同國家落地時所面臨的制度差異、成本結構差距與人才供給瓶頸。這一事件的後續走向，不僅攸關台積電的國際競爭力，也將成為全球各國制定半導體政策時的重要參考。

TSMC’s semiconductor plant in Arizona has recently become the center of attention in the global chip industry due to a dramatic collapse in profitability. According to the latest financial data, the factory’s earnings plunged from NT$4.232 billion in the second quarter of 2025 to only NT$41 million in the third quarter—a staggering 99% decline. This near-total wipeout of profit has raised serious doubts about whether TSMC’s investment of more than NT$1.2 trillion in the United States can realistically be recovered in the long term. It also intensifies scrutiny over whether the company’s high-profile U.S. expansion may be turning into a costly burden.

TSMC’s decision to build fabs in Arizona was originally intended to support Washington’s policy of bringing semiconductor manufacturing back to American soil, while simultaneously strengthening the resilience of TSMC’s global supply chain. Yet the practical challenges of operating in the U.S. have proven far more severe than anticipated. Construction delays, cost overruns, a shortage of local high-skilled semiconductor labor, and complications stemming from U.S. labor regulations have all contributed to mounting financial pressure on the project.

The primary reason behind the collapse in profitability lies in the accelerated transition of the Arizona Fab 2 toward 3-nanometer and other advanced manufacturing processes. Unlike Fab 1—which focuses on mature nodes and operates relatively smoothly—Fab 2 is designed to supply the soaring demand for cutting-edge chips driven by the AI boom. Moving into advanced nodes requires exceptionally expensive manufacturing tools, complex process technologies, and highly specialized engineering talent. These factors have pushed capital expenditures to extreme levels and sharply increased operating costs, directly eroding the plant’s ability to generate profit.

Beyond the transition to advanced nodes, the fundamental cost structure of manufacturing chips in the United States is inherently unfavorable. Construction, equipment installation, insurance, facility maintenance, and even basic operating expenses are significantly higher than in Taiwan, Japan, or South Korea. The local talent pool cannot fully support leading-edge semiconductor production, forcing TSMC to rely on a large number of expatriate engineers, which further inflates labor and training costs. These structural disadvantages cannot be solved quickly, suggesting that TSMC’s U.S. operations may continue to lag behind its facilities elsewhere in terms of profitability for years to come.

While the American market remains strategically important to TSMC, current financial results indicate that the Arizona project is far from reaching the efficiencies the company enjoys in East Asia. With continued investments required for 3-nanometer and even more advanced nodes, financial recovery is unlikely in the short term. Analysts and industry observers are increasingly questioning whether TSMC should reassess its U.S. strategy to prevent these massive investments from becoming a long-term drag on the company’s finances.

Overall, the profit collapse at the Arizona plant highlights not just a single company’s operational challenge but a broader structural dilemma facing the global semiconductor industry. It reveals the stark differences in cost structures, regulatory environments, and talent availability across countries attempting to localize chip production. How TSMC navigates this crisis will have implications far beyond its own competitiveness and may influence how governments worldwide shape their semiconductor industrial policies in the coming decade.