AI infrastructure has exposed a fragile materials supply chain. The industry’s response must extend beyond purchasing and into design, qualification and long-term capacity planning.
By Doug Dixon
ABSTRACT
A tightening supply of electronic-grade glass cloth is moving from a specialist materials issue to a strategic constraint for printed circuit board and semiconductor-package production. Glass cloth is the woven fiberglass fabric that reinforces almost every rigid PCB and IC-package substrate. Demand for its highest-performing grades, low-CTE glass that resists thermal expansion and low-loss glass that limits high-frequency signal loss, is rising faster than qualified capacity. These grades feed AI accelerators, servers and high-speed networking. The immediate results are higher laminate prices, longer lead times and material allocation. The larger risk is that OEMs treat the problem as a temporary purchasing disruption. Industry leaders should instead establish approved material alternatives, bring fabricators into stackup decisions earlier and recognize that changing the glass construction can affect signal integrity, layer-to-layer registration, warpage and long-term reliability.
A small material with an industrywide impact
The electronics industry has spent the past several years talking about shortages of semiconductors, memory, advanced packaging capacity and skilled labor. The next constraint is less visible but just as consequential: electronic-grade glass cloth, the woven fiberglass reinforcement used in copper-clad laminates and semiconductor package substrates. It is the structural skeleton inside the board, and its properties help set both mechanical stability and electrical performance.
This is not a generalized shortage of ordinary fiberglass. The pressure is concentrated on specialty woven glass, including low-CTE T-glass and low-dielectric, or low-Dk, glass used in advanced substrates and high-speed printed circuit boards. Low-CTE means a low coefficient of thermal expansion, so the material grows and shrinks very little as it heats and cools, which keeps large, dense packages dimensionally stable. Low-Dk glass carries high-frequency signals with less distortion and loss. These constructions provide dimensional stability, mechanical reinforcement and controlled electrical performance. They are difficult to manufacture consistently, and the supply base is concentrated among a small number of qualified producers.
That distinction matters. Calling the situation a fiberglass shortage makes it sound like a commodity problem that can be solved by buying more material from another mill. It cannot. Glass yarn diameter, weave geometry, surface treatment, cleanliness, thickness and dielectric characteristics all must stay within narrow limits. The cloth then has to be qualified, meaning tested and formally approved, by the laminate producer, the PCB fabricator and, in many cases, the OEM. Announced capacity is not the same as qualified capacity, and only qualified capacity can actually ship into a released design.
AI demand has changed the allocation equation
The build-out of AI infrastructure is accelerating demand for advanced glass constructions. AI accelerators and the networking hardware around them use large packages, high layer counts, low-loss materials and physically large boards. New rack architectures also require more complex backplanes, midplanes and high-speed interconnect. Each of these systems consumes far more premium laminate and specialty glass than a conventional enterprise server.
The effects will not remain confined to AI products. Glass weavers and laminate manufacturers naturally prioritize materials with the strongest demand and margins. Suppliers are restricting the availability of standard E-glass as production shifts toward low-Dk glass yarns and fabrics.
This combination further tightens open-market supply. As capacity moves toward advanced materials, fewer standard constructions remain available. The resulting constraint may begin with specialty glass fabric but can cascade into high-Tg laminates, halogen-free systems, thin glass styles and commonly specified FR-4 materials.
The first symptoms are already visible
Price and lead-time pressures are already moving through the laminate supply chain. Industry reports indicate that advanced-laminate lead times have approached 140 days, while some standard FR-4 lead times have increased from a few days to several weeks. In certain cases, typical lead times of eight to 12 weeks have extended to 20 to 30 weeks. South Korean copper-clad laminate import prices also reportedly reached record levels during the first half of 2026.
Lead time may become a bigger problem than price. A purchasing team can sometimes absorb a material increase. It cannot recover a product launch when the exact prepreg, core thickness or glass style the board’s layer stackup, the defined arrangement of copper layers, cores and prepregs, requires is unavailable for three months. Quick-turn prototype shops are especially exposed, because their business model assumes immediate access to a broad material inventory.
Material substitution is an engineering decision
The predictable response to a shortage is to ask for an equivalent laminate. That phrase can hide substantial technical risk. Two laminate systems marketed for similar applications can differ in dielectric constant and loss across frequency, Z-axis expansion, resin flow, copper adhesion, moisture absorption, resistance to conductive anodic filament formation and drilling behavior. Conductive anodic filament, or CAF, is a failure mode in which a conductive path grows along the glass fibers between adjacent features and eventually shorts them.
The glass construction itself affects the effective dielectric constant, impedance and propagation delay of the finished board. In very-high-speed differential channels, a change in weave style or resin-to-glass ratio can alter insertion loss, the signal energy lost along a trace, and worsen fiber-weave skew, a timing mismatch that appears when the two lines of a differential pair sit over different amounts of glass and resin. The substituted construction may require new field-solver modeling of the transmission lines, adjusted trace widths and spacing, updated impedance test coupons and additional validation.
The mechanical effects matter just as much. Low-CTE glass controls dimensional movement and warpage in large packages and multilayer structures. Moving to a higher-CTE construction can reduce layer-to-layer registration margin, the tolerance that keeps features on different layers aligned, increase stress on microvias and plated through-holes, and complicate fine-pitch assembly. In aerospace, defense, automotive and medical work, a substitute can also trigger new reliability testing, customer approval and documentation changes.
For these reasons, purchasing should never approve a laminate substitution on its own. Materials engineering, signal integrity, PCB design, fabrication and reliability all need to evaluate the complete construction. An alternate material is not truly approved until its electrical, mechanical, fabrication and compliance implications are understood.
Leadership must move upstream
The strategic mistake would be to manage this shortage only through expediting and supplier pressure. No amount of escalation creates qualified glass capacity. Leaders should instead move material-risk decisions earlier in product development.
First, OEMs should identify every program that depends on a single proprietary laminate or a tightly constrained glass construction. The review should capture core and prepreg thicknesses, resin content, glass style, copper profile and the performance values that actually drive the selection. If a requirement is essential, document it. If it is merely inherited from an earlier design, challenge it.
Second, approve alternatives at the stackup level. A generic note such as “FR-4 or equivalent” offers little protection. The fabricator needs a defined electrical and mechanical performance envelope, along with permission to propose constructions it can source consistently. The OEM then needs a controlled process for simulation, coupon validation and reliability approval.
Third, involve the PCB fabricator before the layout is frozen. Fabricators know which laminates and glass styles are in inventory, which substitutes are already qualified and which constructions can be built across multiple factories. Early engagement can let a design team hold its impedance, thickness and reliability targets while using a more available material set.
Fourth, forecasting has to become more credible. Suppliers cannot justify new furnaces, weaving equipment and treatment lines on the strength of emergency purchase orders. Better demand visibility, longer-term agreements and realistic allocation commitments will be needed to support expansion without recreating the severe overcapacity cycles that have historically discouraged investment.
Finally, commercial teams should make material risk visible in customer commitments. Quotations should identify the laminate family, the assumptions behind availability and the period for which pricing stays valid. Supply agreements should define how alternates are proposed, who approves them and which verification steps are required. This is not an invitation to bury customers in exclusions. It is a way to keep technical decisions from being made under schedule pressure after an order is placed. Transparent communication also gives customers what they need to place forecasts, fund safety stock where it makes sense and decide which programs get priority when allocation hits.
A chance to strengthen the supply chain
The glass-cloth shortage is likely to persist until new production is installed and qualified. Current market estimates point to meaningful relief in 2027, but schedules can slip, and initial output will be claimed quickly by large strategic customers. The industry should plan for continued constraint rather than bet product schedules on a precise recovery date.
The pressure also creates an opening. It can force better communication among material suppliers, fabricators, EMS providers and OEM engineering teams. It can accelerate qualification of second sources, improve material traceability and eliminate specifications that are narrower than the product actually needs. It is also pushing genuine innovation in the reinforcement itself.
Glass cloth sits deep in the supply chain, but its properties reach all the way to system performance and field reliability. The companies that understand that connection will protect their schedules without compromising the product. Those that treat the shortage as a purchasing inconvenience may find out too late that the material choice was baked into the design.
The leadership question is straightforward. Will we wait for allocation notices, or will we design supply resilience into the product? The current shortage is giving the PCB industry a chance to answer before the next constrained material arrives.
Doug will be attending PCB West Tradeshow the week of Sept 28 – Oct 2nd, along with Ventec and other key industry players.
Doug Dixon is CEO and CMO of 360 Circuits, a marketing agency serving the semiconductor, printed circuit board fabrication and assembly industries.