The rapid development of artificial intelligence is driving a sharp increase in the demand for the entire printed circuit board (PCB) material chain, with significant rises in both price and output. According to the “Electronic Times Asia,” supply pressures are no longer limited to low thermal expansion coefficient (CTE) glass fiber fabrics used for CoWoS substrates; shortages are spreading to low dielectric constant (Dk) glass fiber fabrics for copper-clad laminates (CCL), high-capacity low-profile (HVLP) copper foil, glass substrates with high thermal conductivity, and coating micro-drills required for PCB drilling.
As the requirements for PCB materials increase due to AI hardware, the issues are becoming more severe—higher thermal performance substrates, advanced copper foils, and specialized resin systems are needed, all of which take months to meet qualification standards and achieve mass production. Material suppliers report that order backlogs have extended to 20-30 weeks, whereas the typical delivery cycle was only 8-12 weeks two years ago.
Industry insiders point out that the next generation of AI server construction by GPU and ASIC suppliers is roughly planned for the second half of 2026. If upstream PCB material capacities cannot be expanded in time, the shortage of advanced glass fiber fabrics may hinder the production of CoWoS substrates.
The “Electronic Times” previously reported that most of the high-end glass fiber capacity of Japan’s Nitto Denko has been booked by Nvidia and other GPU manufacturers, forcing ASIC developers to turn to Taiwanese suppliers, including FFG and Taiwan Glass Industry Corporation, for more reliable supply.
Glass substrates and their hybrid substrates are also highly competitive alternatives to traditional CoWoS.
In-Depth Analysis of Bottlenecks
PCB drilling tools are also facing supply shortages. Topoint Technology’s global market share is approaching that of Japan’s Union Tool, which states that 2025 will be the first year for large-scale production of AI server packaging substrates and complex multilayer boards. The demand for advanced drill bits is accelerating, with their aspect ratios expected to rise from 20 times to 30-40 times.
Topoint’s president, Karen Lin, noted that the number of layers in PCB for AI servers is significantly increasing, from a minimum of 20 layers to a maximum of 90 layers. A drill bit that could originally process two or three circuit boards now only needs to process one, effectively doubling or tripling the demand and causing order congestion.
She added that as CCL shifts to M9 specifications, some solutions are adopting quartz fiber, which increases hardness and accelerates drill wear. The shortened tool life requires advanced coatings to enhance durability and chip removal performance. Currently, only Topoint and Union Tool can supply drill bits of this level, highlighting the risks of concentrated supply.
Copper Foil Embrittlement Composite Challenges
HVLP copper foil is becoming another bottleneck. As the industry upgrades from HVLP2 to HVLP4, even though companies like Mitsui Mining & Smelting, Furukawa Electric, Luxembourg’s Circuit Foil, and Taiwan’s Kinsus Interconnect Technology are planning to expand production, by 2026, their capacity will only meet half of the demand for AI servers. The delivery cycles for these specialty foils are expected to be significantly extended, threatening the delivery schedule of PCBs.
PCB industry executives report that since the beginning of 2025, the shortages of glass fiber fabrics, drill bits, and copper foil have intensified. Although suppliers are racing to expand advanced production lines, capacity still lags behind demand. This imbalance may create opportunities for second-tier suppliers, potentially reshaping the upstream AI server materials market.
Advanced Applications: AI-Driven Supply Chain Resilience Strategies
Leading electronics manufacturers are deploying AI-based supply chain optimization systems to address the shortage of PCB materials. These systems analyze historical demand patterns, supplier capacity constraints, and geopolitical risks to optimize material allocation across product lines. According to research by Gartner, AI-based supply planning can reduce forecast errors by 20-50% in volatile market conditions.
The most complex approaches involve dynamic supplier scoring, which combines real-time capacity data, financial stability indicators, and geopolitical risk factors. Companies adopting this approach maintain 2-3 qualified suppliers in each critical material category, rather than relying on single-source relationships.
Future Trends: Capacity Expansion and Demand Realities
Industry analysts predict that by 2026, the demand for PCB materials will continue to exceed capacity expansion. New production facilities require 18-24 months to reach full operational capacity, while the demand for AI hardware shows no signs of slowing down. The U.S. Department of Commerce’s “Chips and Science Act” includes provisions for domestic PCB substrate production, but these facilities will not reach effective capacity until 2027-2028.
Meanwhile, as China controls a significant portion of the PCB materials value chain, geopolitical tensions have also exacerbated the complexity. Supply chain leaders must balance cost optimization with supply security—this requires complex risk modeling and scenario planning capabilities.
Proactive Planning Determines Success
PCB material shortages are not just a temporary supply issue, but a fundamental shift that requires new supply chain strategies. Companies investing in real-time supplier visibility, diversified sourcing, and AI-driven demand planning will maintain a competitive edge, while others will face production delays and cost fluctuations.
Feedback Channel: This article overview. If you have article submissions, interviews, or leading technologies or product solutions, please contact the deputy editor of Future Semiconductor, Qi Daozhang (VX: 19910725014, join the advanced packaging industry chain group).