After years of powering ahead as NVIDIA’s (NVDA) premier “shadow stock” in Asia, Taiwan Semiconductor Manufacturing Company (TSMC) is now increasingly vying for investor attention with other AI-related stocks. As the mainstream adoption of AI drives demand for more hardware beyond NVIDIA’s high-end chips, traders are pursuing a broader range of beneficiaries. From tight supply in memory chips to advancements in robotics, multiple AI sub-themes are attracting capital inflows. At the same time, single-stock position limits are prompting funds to diversify. Retail investors, who have long been familiar with TSMC through its American Depositary Receipts (ADRs), now have a wider array of Asian tech stock alternatives.
As a result, TSMC’s stock this year is underperforming that of chip designer MediaTek by the widest margin since 2009. Meanwhile, the market capitalization gap between TSMC and Samsung Electronics, the world’s largest memory chipmaker, has noticeably narrowed, with Samsung having joined the “$1 trillion market cap club.” Jason Hsu, Chief Investment Officer at Rayliant Global Advisors, notes that the market is witnessing a “structural shift away from TSMC toward other companies,” with newly raised fund capital flowing disproportionately into other tech firms that are also benefiting from record AI capital expenditure.
Although TSMC’s stock has risen 44% this year on strong sales and earnings, both MediaTek and Samsung Electronics have seen their shares gain nearly 150%. This situation mirrors that of NVIDIA—despite its continued robust growth, its stock performance has begun to lag behind some AI-related stocks.
There is little doubt that TSMC, as the manufacturer of nearly all of NVIDIA’s advanced GPUs, will continue to benefit from AI demand for some time. However, as AI increasingly moves into the “inference” phase—focused on specific tasks—other players are starting to emerge. Hot trades are appearing among chip design firms; for example, MediaTek is assisting Google in developing application-specific integrated circuits (ASICs). At the same time, demand is growing for central processing units (CPUs), which are relatively less technically demanding and can be manufactured not only by TSMC but also by foundries at Samsung and Intel. Brian Ooi, Portfolio Manager at Swiss-Asia Financial Services, states that agentic AI is driving the expansion of the AI trade, as AI agents require more CPUs, and this diffusion trend will continue as AI spending shifts from training to inference.
Furthermore, because TSMC focuses on logic chips and does not directly benefit from the boom in memory and data storage devices, it has missed out on the gains seen in Samsung and its peers’ stocks. Another reason for TSMC’s relative underperformance is that many active funds have a 10% position limit on single stocks. Given that TSMC already accounts for over 40% of the Taiwan Weighted Index, these funds have had to buy other stocks to keep pace with the index’s performance.
