Taiwan Semiconductor Manufacturing Co. (TSMC) reportedly plans to implement a 3–6% increase in chip manufacturing prices from 2027, focusing higher adjustments on advanced node technologies. This move comes as foundry capacity remains stretched and demand from the AI sector continues to grow.
Persistent undersupply characterizes the 2-nanometer and 3-nanometer manufacturing processes, with demand for advanced packaging methods such as CoWoS outpacing available capacity. Processes under 45nm are operating at or near full utilization, and certain 8-inch production lines have exceeded 100% load. Clients with requirements for advanced nodes must secure orders years in advance, with booking commitments now reaching to 2030.
The accelerating adoption of artificial intelligence has broadened demand beyond graphics and specialized processors, now impacting various chip categories, including power management ICs, microcontrollers, networking components, and sensor devices. This cascading demand benefits not only TSMC but also established legacy foundries like UMC, Vanguard International Semiconductor, and PSMC.
For example, Vanguard’s wafer utilization exceeded 90% along with several months of forward order visibility. The complexity of switching suppliers at the advanced node level—due to required design adjustments and qualification processes—has allowed TSMC’s price actions to set a higher overall pricing baseline. Competitors such as Samsung, Intel, UMC, PSMC, and Vanguard are subsequently positioned to adjust prices upward as well.
Rising manufacturing costs have been significantly affected by TSMC’s ongoing international expansion. Constructing fabrication plants in the United States is estimated to cost four to five times more than in Taiwan. Furthermore, TSMC anticipates that the ramp-up for initial 2nm node production could reduce gross margins by 3–4 percentage points, with additional pressure of 2–3 points linked to overseas facility expenses over the next few years.
InnovestX Securities maintains a positive outlook on TSMC and the broader semiconductor foundry sector, supported by TSMC’s strong pricing power and persistent capacity constraints driven by AI demand, particularly across advanced node processes. With 2nm and 3nm capacity remaining tight and order visibility stretching out to 2030, TSMC’s 3–6% price hike reflects its dominant market position and helps offset high overseas expansion costs.
Leading competitors such as Samsung and Intel stand to benefit as elevated market pricing and industry-wide capacity constraints prompt clients to seek alternative production sources. At the same time, legacy foundries including UMC, Vanguard International Semiconductor, and PSMC benefit as expanding AI server infrastructure increases demand for supporting components like PMICs, MCUs, and analog chips, keeping 8-inch capacity tight.
For GPU, AI chip, and custom ASIC designers, InnovestX assesses the net impact as neutral. Leading AI chipmakers like NVIDIA and AMD face higher wafer costs, but their high product value and robust customer demand enable them to pass these cost increases along effectively. Similarly, major custom ASIC developers and key corporate clients like Broadcom, Google, and Apple incur higher manufacturing expenses, but they remain reliant on TSMC because switching advanced foundry providers requires lengthy design and qualification cycles.
Conversely, producers of PMICs and MCUs experience a slightly negative impact, as intense price competition in those segments limits their ability to pass along higher costs. The impact is most severe for designers of consumer chips in smartphones and general electronics, where sluggish end-market demand severely limits pricing power, heightening profit margin risks entering 2027.
