The critical bottleneck in the AI supply chain has shifted from raw materials to the manufacturing equipment needed to process them.
This entire situation began with the explosive growth of AI infrastructure. Tech giants are building massive data centers, which is reflected in NVIDIA’s datacenter revenue more than doubling year-over-year. This buildout requires a massive upgrade in networking speed, with the industry rapidly moving from 400G to 800G and now looking toward 1.6T technology. This isn't just a minor upgrade; it demands fundamentally better and more complex printed circuit boards (PCBs).
To handle these incredible speeds without losing data, you need superior materials. This is where high-performance Copper Clad Laminates (CCLs), specifically advanced grades like M8 and M9, became essential. They form the foundation of the high-layer-count PCBs used in modern AI servers and network switches.
The causal chain leading to today's equipment shortage started with a scarcity of the raw materials needed for these advanced CCLs. First, the supply of specialty low-loss glass cloth, primarily made by Japan's Nittobo, has been tight for years, with new capacity only coming online slowly. Second, copper prices have been volatile due to mine disruptions and trade policy uncertainty, adding another layer of supply chain risk.
Faced with these persistent material shortages, CCL manufacturers like EMC and TUC did something logical: they all tried to secure their future production by launching aggressive, simultaneous expansion plans across multiple countries. To build these new factories, they rushed to place large, early orders for the specialized manufacturing equipment, particularly for prepreg production lines, trying to get ahead of their competitors and the material crunch.
This synchronized rush created the current, more severe bottleneck. The demand for manufacturing equipment overwhelmed suppliers. A key vendor, Asia Metal Industries, now reports its order book is full through the first quarter of 2028. For a CCL maker ordering a new prepreg line today, the wait time is nearly two years. The constraint in the AI supply chain is no longer just about securing glass cloth or copper foil; it's about securing the tools to make the final product, a problem that will likely persist into 2027.
- CCL (Copper Clad Laminate): A base material for printed circuit boards (PCBs), consisting of a laminate sheet clad with copper foil on one or both sides.
- Prepreg: A composite material made from 'pre-impregnated' fibers and a partially cured polymer resin. It is a key component in manufacturing multi-layer PCBs.
- 800G/1.6T Networking: Refers to the data transfer speed of network equipment, measured in gigabits (G) or terabits (T) per second. The move to higher speeds is crucial for AI data centers.