2022 年末生成式AI的快速興起，使全球對高效能運算、資料中心建置等需求大幅增加。基於台灣半導體產業位居全球AI硬體供應鏈的關鍵位置，本研究以台灣上市櫃公司為研究對象，探討生成式AI興起後，AI 半導體供應鏈之附加價值型全要素生產力（TFP）是否相對傳統製造業出現顯著變化。
實證結果顯示，生成式AI出現後，AI 半導體供應鏈之 TFP 相對於合成對照組呈現顯著提升。此結果顯示，生成式AI所帶來的並非僅侷限於如自動化生產或企業內部導入AI等生產層面的影響，同時包含透過模型訓練與推論所需之龐大算力需求，進一步推升AI晶片、先進製程、先進封裝與記憶體等半導體供應鏈之附加價值創造能力。此一能力提升可能來自高附加價值訂單增加、產能利用率提升，以及廠商因應AI需求而加速製程升級與供應鏈整合。相較之下，傳統製造業較難直接受惠於AI硬體需求擴張，且內部導入AI仍需面臨組織調整、人才不足與資本投入限制。
整體而言，生成式AI可能進一步強化台灣半導體產業相對傳統製造業的優勢，並使資本、人才與政策資源配置更加集中。未來政策若無法協助傳統製造業進行數位轉型與人力升級，台灣產業內部的 K 型化發展可能更加明顯。

The rapid emergence of generative AI at the end of 2022 substantially increased global demand for high-performance computing and data center infrastructure. Given Taiwan’s critical position in the global AI hardware supply chain, this study examines publicly listed and over-the-counter firms in Taiwan to investigate whether the value added total factor productivity（TFP）of the AI semiconductor supply chain experienced a significant change relative to traditional manufacturing industries after the emergence of generative AI.
The empirical results indicated that, following the emergence of generative AI, the TFP of the AI semiconductor supply chain increased significantly relative to the synthetic control group. This finding suggests that the impact of generative AI is not limited to production-side effects, such as automated manufacturing or the internal adoption of AI by firms. Rather, the substantial computational requirements associated with model training and inference have also generated strong demand for AI chips, advanced semiconductor processes, advanced packaging, and memory products, thereby enhancing the value-added creation capacity of firms throughout the semiconductor supply chain. This improvement in productivity may stem from an increase in high value-added orders, higher capacity utilization, and accelerated process upgrading and supply chain integration in response to AI-related demand. In contrast, traditional manufacturing industries are less likely to directly benefit from the expansion of AI hardware demand, while their adoption of AI still faces constraints such as organizational adjustment, shortages of skilled labor, and capital investment limitations.
Overall, generative AI may further strengthen the relative advantage of Taiwan’s semiconductor industry over traditional manufacturing industries, leading to a greater concentration of capital, human resources, and policy resources. If future policies fail to facilitate digital transformation and workforce upgrading in traditional manufacturing industries, the K-shaped development within Taiwan’s industrial structure may become even more pronounced.

一、參考文獻
Abadie, A. (2021) . Using synthetic controls: Feasibility, data requirements, and methodological aspects. Journal of Economic Literature, 59 (2) , 391-425.
Abadie, A., Diamond, A., & Hainmueller, J. (2015) . Comparative politics and the synthetic control method. American Journal of Political Science, 59 (2) , 495–510.
Abadie, A., & Gardeazabal, J. (2003) . The economic costs of conflict: A case study of the Basque Country. American Economic Review, 93 (1) , 113–132.
Acemoglu, D. (2025). The simple macroeconomics of AI. Economic Policy, 40(121), 13-58.
Arkhangelsky, D., Athey, S., Hirshberg, D. A., Imbens, G. W., & Wager, S. (2021) .Synthetic difference-in-differences. American Economic Review, 111 (12) , 4088–4118.
Avdiu, B. (2022) . Improving productivity measurement in World Bank Group interventions.World Bank.
Brynjolfsson, E., Rock, D., & Syverson, C. (2018). Artificial intelligence and the modern productivity paradox: A clash of expectations and statistics. In A.Agrawal, J. Gans, & A. Goldfarb (Eds.), The economics of artificial intelligence: An agenda (pp. 23–57). University of Chicago Press.
Bustos, P. (2011) . Trade liberalization, exports, and technology upgrading: Evidence on the impact of MERCOSUR on Argentinian firms. American Economic Review, 101 (1) ,304–340.
Campello, M. (2006) . Debt financing: Does it boost or hurt firm performance in product markets? Journal of Financial Economics, 82 (1) , 135–172.
Campos, N. F., Coricelli, F., & Franceschi, E. (2022) . Institutional integration and productivity growth: Evidence from the 1995 enlargement of the European Union. European Economic Review, 142, 104014.
Caplin, A., Lee, M., Leth-Petersen, S., Sæverud, J., & Shapiro, M. D. (2022). How worker productivity and wages grow with tenure and experience: The firm perspective (NBER Working Paper No. 30342). National Bureau of Economic Research.
Card, D., & Krueger, A. B. (1994). Minimum wages and employment: A case study of the fast-food industry in New Jersey and Pennsylvania. The American Economic Review, 84(4), 772–793.
Corden, W. M., & Neary, J. P. (1982). Booming sector and de-industrialisation in a small open economy. The Economic Journal, 92(368), 825-848.
Goldar, B., Chawla, I., & Behera, S. R. (2020). Trade liberalization and productivity of Indian manufacturing firms. Indian Growth and Development Review, 13(1), 73-98.
Griliches, Z., & Mairesse, J. (1981). Productivity and R and D at the Firm Level. NBER Working Paper No.826. National Bureau of Economic Research.
He, B., Zheng, X., Chen, Y., Li, W., Zhou, Y., Long, X., ... & Zhang, X. (2023). Dxpu: Large-scale disaggregated gpu pools in the datacenter. ACM Transactions on Architecture and Code Optimization, 20(4), 1-23.
Hellsten, M., Khanna, S., Lodefalk, M., & Yakymovych, Y. (2025). The effects of artificial intelligence on jobs: Evidence from an AI subsidy program (IZA Discussion Paper No. 18267).IZA Institute of Labor Economics.
Hoedoafia, M. (2020) . On the link between trade liberalization and firm productivity: Panel data evidence from private firms in Ghana.SSRN.
Hsieh, C. T., & Klenow, P. J. (2009). Misallocation and manufacturing TFP in China and India. The Quarterly journal of economics, 124(4), 1403-1448.
Jovanovic, B. (1982). Selection and the Evolution of Industry. Econometrica, 50(3), 649–670.
Kandilov, Ivan T. & Liu, Xiangping, 2011. "Do Exports Raise Productivity? Plant-level Evidence from the Colombian Agri-food Industries," 2011 Annual Meeting, July 24-26, 2011, Pittsburgh, Pennsylvania 103632, Agricultural and Applied Economics Association.
Kahn-Lang, A., & Lang, K. (2020). The promise and pitfalls of differences-in-differences: Reflections on 16 and pregnant and other applications. Journal of Business & Economic Statistics, 38(3), 613-620.
Lechner, M. (2011). The estimation of causal effects by difference-in-difference methods. Foundations and trends in econometrics, 4(3), 165-224.
Levinsohn, J., & Petrin, A. (2003). Estimating production functions using inputs to control for unobservables. The Review of Economic Studies, 70(2), 317-341.
Lieberman, M. B., & Demeester, L. (1999). Inventory reduction and productivity growth: Linkages in the Japanese automotive industry. Management Science, 45(4), 466-485.
Lileeva, A., & Trefler, D. (2010). Improved access to foreign markets raises plant-level productivity… for some plants. The Quarterly Journal of Economics, 125(3), 1051-1099.
Lin, C. L., Lin, H. L., & Lin, E. S. (2016). Is there a complementary relationship between product and process innovation on productivity in Taiwanese manufacturing firms? Hitotsubashi Journal of Economics, 57(2),139-173.
Luccioni, S., Jernite, Y., & Strubell, E. (2024, June). Power hungry processing: Watts driving the cost of AI deployment? In Proceedings of the 2024 ACM conference on fairness, accountability, and transparency (pp. 85-99).
Melitz, M. J. (2003). The impact of trade on intra‐industry reallocations and aggregate industry productivity. Econometrica, 71(6), 1695-1725.
Murphy, K. M., Shleifer, A., & Vishny, R. W. (1991). The allocation of talent: Implications for growth. The Quarterly Journal of Economics, 106(2), 503-530.
Olley, G. S., & Pakes, A. (1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64(6), 1263–1297.
Petrin, A., Poi, B. P., & Levinsohn, J. (2004). Production function estimation in Stata using inputs to control for unobservables. The Stata Journal, 4(2), 113-123.
Santoni, E., Patriarca, F., & Scarlato, M. (2025). The effects of hiring credits on firms' dynamics: A synthetic difference-in-differences evaluation. The BE Journal of Economic Analysis & Policy, 25(4), 1121-1181.
Sevilla, J., Heim, L., Ho, A., Besiroglu, T., Hobbhahn, M., & Villalobos, P. (2022, July). Compute trends across three eras of machine learning. In 2022 international joint conference on neural networks (IJCNN) (pp. 1-8).IEEE.
二、資料來源
大陸委員會。（2020）。美中科技戰對兩岸科技產業之影響與因應建議。
台灣積體電路製造股份有限公司（2022年4月14日）。2022年第一季法人說明會 [簡報檔案]。https://investor.tsmc.com/chinese/encrypt/files/encrypt_file/reports/2022-04/cf4c59dae5070ba614cc190fc285e5178a384c4c/1Q22Presentation%28C%29.pdf
台灣積體電路製造股份有限公司（2023年3月12日）。2022年度年報。https://investor.tsmc.com/sites/ir/annual-report/2022/2022%20Annual%20Report-%20C.pdf
行政院主計總處。（2025）。110年度產業關聯表部門分類。
李佳臻。（2023年3月10日）。2022年第四季臺灣IC產業回顧與展望。工研院產科國際所。https://ieknet.iek.org.tw/iekrpt/rpt_more.aspx?actiontype=rpt&indu_idno=0&domain=2&rpt_idno=758841064
林育如。（2023年11月23日）。113年研替員額公布！近4千員額，接軌職場「役」馬當先。觀傳媒。https://www.watchmedia01.com/archives/352850
財政部。貿易統計資料查詢。https://web02.mof.gov.tw/njswww/WebMain.aspx?sys=100&funid=defjsptgl
國家發展委員會。（2024年1月）。重要總體統計資料手冊。
經濟部產業發展署。（2025）。「產業創新條例」第10條實施檢討報告。https://www.ida.gov.tw/ctlr?PRO=filepath.DownloadFile&f=executive&t=f&id=19884
資策會產業情報研究所。（2025年5月7日）2025臺灣半導體產值達5.45兆新台幣成長15.4% 7nm以下全球先進製程產能臺廠占全球產能63%。https://mic.iii.org.tw/research.aspx?id=721
資策會數位轉型研究院。（2026年1月29日）。全球產業AI化現況與趨勢觀察。https://ideas-dtri.iii.org.tw/%E7%94%A2%E6%A5%AD%E8%B3%87%E8%A8%8A/ai-6/
僑務電子報僑務新聞網。僑務新聞網頁。https://ocacnews.net/article/300008
Academia Sinica. (2023). 2023 Taiwan economic forecast: A revision resilient consumption amid weak signs of export recovery. Academia Sinica Newsletter. https://newsletter.sinica.edu.tw/en/12320/
Marvell.Custom ASICs. Retrieved June 13, 2026, from https://www.marvell.com/products/custom-asic.html
McKinsey (2024, March 29). Generative AI: The next S-curve for the semiconductor industry? https://www.mckinsey.com/industries/semiconductors/our-insights/generative-ai-the-next-s-curve-for-the-semiconductor-industry
Ministry of Finance.（2025）. Financial and economic analysis for the first half of 2025. https://service.mof.gov.tw/public/Data/statistic/bulletin/114/2025H1%E8%8B%B1%E6%96%87%E5%88%86%E6%9E%90.pdf
OECD.（2025）.The Global Forum on Productivity at 10: Past and future perspectives on reviving productivity growth. OECD Publishing.
OpenAI. ChatGPT. Retrieved June 13, 2026, from https://openai.com/zh-Hant/index/chatgpt/
Semiconductor Industry Association.（2024, February 5）. Global semiconductor sales decrease 8.2% in 2023; market rebounds late in year.https://www.semiconductors.org/global-semiconductor-sales-decrease-8-2-in-2023-market-rebounds-late-in-year/
The Economist. (1977, November 26). The Dutch disease. The Economist, 265(7004), 82–83
TrendForce.（2026年5月6日）。北美業者擴大投資AI資料中心助力，2026年全球九大CSP資本支出上調至8,300億美元。https://www.trendforce.com.tw/presscenter/news/20260506-13032.html
World Semiconductor Trade Statistics. (2025, December 2). Global semiconductor market approaches USD 1 trillion in 2026.https://www.wsts.org/esraCMS/extension/media/f/WST/7310/WSTS_FC-Release-2025_11.pdf.