U.S. AI and semiconductor stocks have recently experienced volatility. Although TSMC announced better-than-expected financial results during its earnings call and revised its 2026 capital expenditure upward to a record high of US$64 billion, lingering market doubts about the AI industry’s subsequent growth momentum have led to a pullback in stock prices, reigniting discussions over a potential AI bubble.
High AI Valuations and Elevated Financing Leverage—Market Volatility Reflects Pressure to Validate Profitability
Hsien-Ming Lien, President of the Chung-Hua Institution for Economic Research (CIER), pointed out that the market’s focus regarding AI industry development has gradually shifted from technological breakthroughs to whether business models can sustainably support massive investments. The central bank recently released a financial stability report identifying four notable risks within the AI industry: high corporate valuations, increased systemic risks from circular trading within the supply chain, rapid expansion in financing scales, and a heavy reliance on off-balance-sheet financing models such as private credit and special purpose vehicles.
Notably, some companies in the AI supply chain simultaneously act as suppliers, customers, investors, and financiers, creating a highly interconnected investment loop. If a key company alters its investment or operational strategy, the impact could be magnified throughout the supply chain and capital markets. Moreover, AI infrastructure has heavily relied on funding sources such as bonds and private credit in recent years. Should the global financial environment tighten, it could drive up industrial financing costs and investment risks.
AI Investment Enters the Next Phase—Business Models and End-User Demand Become Crucial
However, Lien believes that a short-term market correction does not indicate a shift in AI’s long-term developmental trajectory. Rather, the industry is entering a new phase of validating its commercial value. As competition among large language models intensifies with the recent launches of ChatGPT, Gemini, and new contenders from China, the market is increasingly concerned about whether the investments in model development can translate into corporate revenue and profitability.
Lien noted that the biggest challenge currently facing the AI industry is how to convert massive computing power investments into sustainable end-user applications and establish stable business models. If AI technology can continue to deeply integrate into corporate digital transformations, production process improvements, smart manufacturing, and various industrial applications, it will help foster a virtuous cycle that further supports AI infrastructure and related investment demands. Conversely, if end-user demand growth falls short of expectations, the market will inevitably recalibrate its expectations for high-valuation companies.
Taiwan Must Strengthen Its Advanced Manufacturing Advantages and Promote AI Diffusion into Traditional Industries
For Taiwan, the AI wave continues to present significant developmental opportunities, but the more pressing priority is to continuously elevate industrial competitiveness. Lien stated that Taiwan’s economy currently exhibits an imbalance between the technology sector and traditional industries. While tech exports continue to grow thanks to AI demand, some traditional industries still face headwinds. Looking ahead, the government must not only continue supporting the semiconductor and AI supply chains but also assist more traditional industries in adopting AI technologies. This will improve production efficiency and increase the value of products, thus accelerating industrial transformation and modernization.
As global AI competition continues to heat up, Taiwan must solidify its core advantages in advanced manufacturing processes, high-end packaging, and AI servers. It should also drive the diffusion of AI applications into manufacturing, services, and SMEs. These efforts will ensure that AI gradually transitions from a mere capital market narrative into a new engine driving overall industrial upgrading and economic growth.
Author: CIER Editorial
Team Date: July 21, 2026