Glass Fiber Cloth Shortage Creates Critical AI Hardware Bottleneck, Squeezing Supply Until Mid-2027

A shortage of special glass fiber cloth, a critical material for AI server circuit boards, has created a major supply chain bottleneck.

This is caused by a surge in demand from new AI hardware and a highly concentrated supply, dominated by the Japanese company Nittobo.

The situation is expected to increase AI hardware costs and potentially delay deployments until at least mid-2027, when new production capacity is anticipated.

A surprisingly low-tech material, glass fiber cloth, has become a critical bottleneck threatening the pace of the entire AI hardware revolution.

At its core, this special cloth is a key ingredient for the high-performance Printed Circuit Boards (PCBs) that act as the nervous system for AI servers. Think of it as the ultra-strong, lightweight skeleton within the green circuit boards, ensuring signals travel at incredible speeds without interference. As AI models become more powerful, the demand for boards that can handle faster data transfer has skyrocketed, and this requires very specific, high-quality glass cloth.

So, what caused this sudden shortage? There are three main factors at play.

First is a massive demand shock. Companies like NVIDIA are releasing next-generation AI platforms, such as the Rubin and Groq LPX rack systems. These new systems use larger, more complex circuit boards to connect hundreds of processors, significantly increasing the amount of high-grade glass cloth needed per server rack. The push towards faster networking speeds like 800G and 1.6T also demands these premium materials.

Second, the supply is extremely concentrated. A single Japanese company, Nitto Boseki (Nittobo), dominates the market for the highest-quality 'T-glass' and 'NER-glass', controlling up to 90% of the supply. Building new factories to produce this material is a slow and complex process that takes years to perfect. Even with government support and planned expansions, significant new supply isn't expected to be ready and reliable until mid-2027.

Finally, this bottleneck is compounded by rising costs of other raw materials. The price of copper, used for the conductive layers of PCBs, has been near record highs. When the cost of copper foil, resin, and now glass cloth all increase simultaneously, the price of the finished circuit board inevitably goes up.

This translates to real-world consequences. The cost of a finished PCB could increase by 5-10% or more, a significant figure that gets passed on to cloud providers and eventually consumers. More importantly, it could slow down the deployment of new AI data centers, creating a physical limit on AI's growth until this humble material's supply chain catches up.

Glossary:
PCB (Printed Circuit Board): The green board that electronically connects components like chips and memory in devices from smartphones to AI servers. It's the foundation of all modern electronics.
Low-Dk/Df glass cloth: A specialized type of glass fiber cloth with low dielectric constant (Dk) and dissipation factor (Df). These properties are essential for minimizing signal loss at very high frequencies, making it crucial for AI and high-speed networking hardware.
Nittobo (Nitto Boseki): A Japanese company that holds a near-monopoly on the production of high-performance glass fiber cloth required for advanced PCBs used in AI infrastructure.