Artificial intelligence's explosive demand for computing power is pushing the global semiconductor supply chain into unprecedented territory.

According to reports from September 3rd, senior executives at Taiwan Semiconductor Manufacturing revealed that equipment demand has soared from a baseline level to 1.9 times its original value between the end of last year and July of this year, an increase of approximately 90% in less than eight months. Even as the company embarks on its most aggressive expansion ever, simultaneously building nearly twenty fabrication plants worldwide, capacity growth is struggling to keep pace with demand.

Simultaneously, this AI-driven surge is also cascading into areas like substrates, exacerbating supply chain imbalances. Statements from leadership at Taiwan Semiconductor Manufacturing and Unimicron indicate that the strain from AI's computational expansion has moved beyond advanced processing nodes to encompass advanced packaging, substrates, and related materials, forcing a deep-seated restructuring across the entire semiconductor ecosystem.

Equipment demand nearly doubles in half a year, yet capacity remains insufficient

Cliff Hou, Senior Vice President and co-Chief Operating Officer at Taiwan Semiconductor Manufacturing, stated at a recent event that in his three decades in the industry, he has never witnessed demand growing at such speed and frequency. Data shows the company's equipment demand index climbed from 1x at the end of last year to 1.9x by July, a growth of about 90% in just over six months.

To address this pressure, Taiwan Semiconductor Manufacturing is expanding capacity on an unprecedented scale. Currently, the company is simultaneously constructing 13 wafer fabs in Taiwan and another five to six in other global regions, totaling roughly 20 sites. This is a significant jump from the typical four to five fabs under construction in previous comparable periods. Despite this massive scale-up, Hou admitted that it is still not enough to satisfy demand.

Hou pointed out that AI's relentless pursuit of computational power and energy efficiency is accelerating upgrades in advanced process technology, advanced packaging, and system-level performance. Unlike the relatively linear production models of the past, today's supply chain must not only ensure products move smoothly to the next manufacturing stage but also consider performance matching, manufacturability, and yield rates between upstream and downstream processes, placing higher demands on supply chain coordination.

Substrate supply-demand imbalance worsens, key upstream suppliers become bottleneck

The AI boom is also significantly impacting the substrate segment. James Chien, Chairman of Unimicron, stated that the supply-demand imbalance in the substrate market is worsening. Larger dimensions, higher layer counts, and more complex substrate designs are continuously raising manufacturing challenges, including increasingly stringent requirements for lower coefficients of thermal expansion (CTE) and dielectric constants.

Chien noted that the substrate supply chain is at a critical inflection point, with core production tools and materials like ABF film and TGV-related technologies heavily dependent on small and medium-sized Japanese suppliers. Over the past 18 months, Unimicron has held more than ten in-depth meetings with clients and Japanese vendors to coordinate responses to the mounting supply chain pressures.

Importantly, the demand from AI is not solely concentrated in advanced process nodes. Chien believes that advanced and mature technologies are highly complementary within the AI ecosystem, and emerging AI applications will create new market opportunities for mature nodes, particularly in customized peripherals and sensor applications.

He specifically highlighted areas such as high-density bonding, specialty memories, logic-based chiplets, silicon interposers with deep trench capacitors, and silicon photonics as fields with significant growth potential at mature technology nodes. In essence, the expansion of AI computing power is driving comprehensive advancements across the entire spectrum, from advanced processing to packaging, substrates, materials, and mature nodes.