How will the competitive landscape of the fiberglass industry evolve, constrained by both technology and resources?
Demand for Specialty Electronic Fabrics is Upward; Q Fabric Volume Release Timing Becomes Key
1. LowDk-2/LowCTE: Supply-Demand Gap Supports Continued High Prosperity
The explosive growth in demand for AI servers and advanced packaging is driving exponential growth in demand for high-end electronic fabrics.
NVIDIA's GB200 architecture AI servers use 5-8 times more PCBs per unit than traditional servers. Coupled with the global eight major cloud vendors' planned capital expenditure of over $420 billion by 2025 (35% of which will be invested in AI computing infrastructure), the supply bottleneck for LowDk-2 and LowCTE electronic fabrics continues to become increasingly prominent.
Currently, the global high-end electronic fabric market has long been monopolized by Japanese manufacturers such as Nittobo and Asahi Kasei, with a combined market share exceeding 70%. Nittobo's capacity utilization rate has reached over 95%, with expansion cycles as long as 18-24 months and a cautious expansion intention, mainly due to insufficient supply of high-end kiln equipment. In the second quarter of 2025, the global shortage of high-end electronic fabrics reached 25%-30%, with prices rising by 250%-300% cumulatively since the beginning of 2024. Specifically, the unit price of LowDk-2 electronic fabric rose from 30 yuan/meter to 105 yuan/meter, and the unit price of LowCTE electronic fabric exceeded 120 yuan/meter, both still in a state of supply shortage.
Our calculations indicate that the global supply shortage of LowDk-2 and LowCTE electronic fabrics will continue to vary in 2026-2027. The shortage is expected to be approximately 120 million meters in 2026, narrowing to 80 million meters in 2027, but still remaining. Coupled with Japanese manufacturers' capacity expansion lagging behind demand growth (Nittobo only plans a 10% expansion in 2026), domestic mass-production enterprises will continue to benefit from both increased volume and price.
2. Q-Fabric: Demand Surge Depends on Downstream Validation; 2026 to be a Key Window
Q-Fabric, a core bottleneck material for NVIDIA's Rubin architecture AI servers, boasts a lower dielectric constant (2.3 compared to LowDk-2, approximately 3.4) and a loss factor (Df) of ≤0.003, effectively reducing signal transmission loss and supporting ultra-high-speed 224Gbps signal transmission. It is a core material for next-generation high-end PCBs, widely used in AI server motherboards, high-speed switch backplanes, and other applications. Our estimates project global Q-Fabric demand to be 16.85 million and 71.88 million meters in 2026 and 2027, respectively. The key prerequisite for large-scale demand surge is the mass production of next-generation GPUs (such as subsequent iterations of NVIDIA Hopper) and Rubin architecture servers. A single Rubin architecture server uses approximately 2.5 square meters of Q-Fabric, 1.8 times that of the GB200 architecture.
Currently, Q-type fabric is in the small-batch testing and terminal certification stage with CCL customers. Leading CCL manufacturers such as Taikoo Electronics and Unimicron Technology have completed small-batch trials. The first quarter of 2026 is a critical period for the implementation of terminal verification results. If verification is successful, mass production will gradually begin, with large-scale production expected in the second quarter (monthly capacity exceeding 1 million meters). As of early 2026, the price of Q-type fabric has risen from 200 yuan/meter in the fourth quarter of 2025 to 300 yuan/meter, an increase of over 50%, and is expected to have another 30% increase for the whole year. If shipments of Rubin architecture and 1.6T switches grow rapidly (1.6T switch shipments are expected to increase by 200% year-on-year in 2026), it will directly drive Q-type fabric demand to exceed expectations, with total demand potentially reaching over 20 million meters for the year.
Q-grade fabric's core barriers to entry are prominent: Yield and resource-driven cost competitiveness.
1. Production Process: The drawing stage is the core barrier, and a complete industry chain layout is rare.
Unlike mature glass fiber manufacturing processes, Q-grade fabric production requires a complete process of "high-purity quartz sand - rod making - drawing - twisting - warping - weaving - post-treatment." The core barrier is concentrated in the drawing stage. Electronic-grade quartz fiber requires high-purity quartz sand with SiO₂ purity ≥ 99.998% (4N8). Forming relies on high-temperature melting above 1800℃ and precision drawing processes. During the drawing process, the fiber diameter must be controlled at 8-10μm, and the single filament breaking strength must reach above 3.5GPa. This places extremely high demands on energy efficiency and the yield of self-developed equipment.
Currently, only a few domestic companies, such as Feilihua, have achieved integrated production across the entire industry chain. Through their self-developed gas refining and fiber drawing equipment, they have increased the yield of Q-cloth to over 90%, surpassing Japanese manufacturers (approximately 87%) by 3 percentage points. The localization of core equipment compared to imported equipment helps reduce costs by 30%, and their gross profit margin is 15-20 percentage points higher than the industry average.
2. Upstream Resources: Tight Supply of High-Purity Quartz Sand Constrains Industry Expansion
High-purity quartz sand is the core raw material for Q-cloth production, accounting for over 40% of production costs. Globally, high-purity quartz resources of 4N8 and above are almost monopolized by the Spruance mine in the United States, accounting for over 90% of global supply. China relies heavily on imports for 4N5 and above high-purity resources, with import prices reaching as high as 30,000 yuan/ton (compared to only 3,000 yuan/ton for ordinary photovoltaic quartz sand).
Meanwhile, there is a severe overcapacity in low-end photovoltaic quartz sand. In 2025, the domestic photovoltaic quartz sand capacity utilization rate is expected to be less than 40%, with the entire industry chain operating at a loss, creating a polarized pattern of "low-end oversupply and high-end shortage." Currently, only a few domestic companies, such as Feilihua and Quartz Shares, have broken through the 4N8 high-purity quartz sand purification technology. However, their production capacity is limited, further raising the entry barrier to the Q-fiber fabric industry. Companies with stable high-purity quartz sand supply channels or independent purification capabilities will have a significant first-mover advantage.
3. Customer Certification: Long Cycle, High Barriers, Binding Core Customers is Key
The downstream end customers of Q-fiber fabric are mainly technology giants such as NVIDIA, TSMC, and Intel. These customers have extremely high requirements for the supplier's technical stability, production capacity assurance capabilities, and product consistency. The certification cycle is as long as 12-24 months, and once certified, a long-term lock-in mechanism is formed, making replacement extremely costly (end customers need to invest an additional 6-8 months of verification period to change suppliers). Currently, Feilihua's Q-cloth products are in the small-batch testing and terminal certification stage with clients, and have not yet passed the formal certification of NVIDIA and Taiguang Electronics. Production capacity needs to be secured after certification is finalized. Taishan Fiberglass, a subsidiary of Sinoma Science & Technology, has entered the customer certification stage and achieved mass production. Its special fiber cloth products have passed the certification of leading CCL manufacturers such as Taiguang Electronics and will ultimately be used in NVIDIA-related equipment, deeply binding to the core needs of the AI industry chain. Honghe Technology is actively promoting the terminal certification of its Q-cloth, with the industry generally expecting the second quarter of 2026 to be a key window for the certification results. Leading companies are building secondary barriers by leveraging their early certification progress and technological accumulation, further widening the gap with industry followers.
Competitive Landscape: Short-term leaders remain stable; long-term differentiation into two types of core companies
1. Short-term Landscape: Monopoly unlikely to change, domestic substitution accelerating breakthrough
Japanese manufacturers dominate the global high-end electronic cloth market. Nittobo has a first-mover advantage in the LowDk-2 and Q-cloth high-end markets, while Asahi Kasei focuses on the LowCTE sub-segment. The leading Japanese manufacturers collectively hold over 70% market share. While domestic companies are actively developing Q-fiber fabric projects, such as Lighttech Optoelectronics, which has announced the development, production, and sales of Q-fiber fabric through its subsidiary Lighttech Quasi-Stone, currently in the initial preparation stage, having completed the formation of its core team and acquired some production equipment, it is advancing capacity expansion and engaging with upstream and downstream partners in the industry chain.
Overall, most domestic companies involved in this sector can only achieve partial breakthroughs in non-core areas such as rod manufacturing and fabric weaving, struggling to overcome the dual barriers of wire drawing technology and end-customer certification. The competitive landscape is unlikely to change in the short term. Benefiting from the booming AI-PCB market, leading domestic companies are accelerating domestic substitution. Companies like Honghe Technology and Feilihua have entered the global core supply chain. By 2025, the market share of domestic companies in the global high-end electronic fabric market is expected to increase to 22%, an 8 percentage point increase from 2023, with the domestic market share showing a rapid upward trend.
2. Long-Term Landscape: Two Types of Enterprises Dominate, Differentiated Competition Becomes the Mainstream
In the medium to long term, the Q-fiber fabric industry will see the emergence of two types of core competitive enterprises, presenting a differentiated competitive landscape: First, vertically integrated enterprises across the entire industry chain, represented by Feilihua, which, through the construction of high-purity quartz sand production capacity and the layout of quartz electronic yarn production lines, are gradually achieving independent control over the entire process from "quartz sand purification to fiber drawing to post-weaving processing," establishing an advantage in the high-end market based on technological accumulation.
Second, enterprises specializing in core segments, represented by Honghe Technology, are focusing on niche areas such as ultra-thin and extra-thin fabrics. Leveraging their globally leading market share and LowDk-2 mass production technology accumulation, they are building a moat in the high-end niche market, while simultaneously rapidly advancing Q-fiber fabric certification and mass production through deep cooperation with leading CCL manufacturers. In addition, some enterprises may gain a cost advantage by securing upstream high-purity quartz sand resources, thus occupying the mid-range Q-fiber fabric market.
From a capacity layout perspective, 2026 is a peak period for the release of Q-fiber fabric capacity: Feilihua's wholly-owned subsidiary, Hubei Dingyi New Materials, is constructing a 1,000-ton capacity quartz electronic yarn production line (corresponding to approximately 1.2 million meters/year of Q-fiber fabric capacity). The pace of capacity release is highly aligned with the progress of end-user certification and the mass production of Rubin architecture servers. Sinoma Science & Technology's 35 million meters/year special fabric project (including Q-fiber fabric-related capacity) is planned to reach full production in the second quarter of 2026. Leveraging the advantages of central state-owned enterprise resources and existing cooperation with CCL customers, it possesses competitiveness in supply chain security. Light Optoelectronics, through its holding subsidiary Light Quash, is conducting Q-fiber fabric business. Currently in the initial preparation stage, it has completed the formation of its core team and is promoting capacity construction and liaising with upstream and downstream partners in the industry chain. There is currently no clear timetable for capacity implementation. Overall, the effective global Q-fiber fabric capacity (yield ≥80%) in 2026 is approximately 25 million meters, which will still be insufficient to fully cover demand, and the tight supply-demand balance will continue.
Investment Insights and Risk Warnings
Core Investment Logic
1. In the short term, focus on Honghe Technology, a leading manufacturer of LowDk-2/LowCTE products. Its high-end products have passed customer certification and are shipping stably. Benefiting from supply-demand gaps and price increases, its performance is highly elastic. Its Q-cloth certification progress is also worth monitoring.
2. In the medium to long term, focus on Feilihua, a potential leader in the entire Q-cloth industry chain. It has a complete supply chain layout from quartz sand to electronic yarn to electronic cloth. End-user certification progress is the core catalyst, and its growth potential is significant after capacity release.
3. Pay attention to Sinoma Science & Technology, a company with high capacity elasticity. It has a complete layout of all types of specialty fabrics, and its products have entered NVIDIA's supply chain. After its 35 million meters/year project reaches full production, it will rank among the top in the industry.
4. Track potential companies in specific sub-sectors, such as Quartz Shares (advantages in high-purity quartz sand) and Light Optoelectronics (Q-cloth business preparation and capacity construction progress).
Risk Warnings
1. Risk of Unexpected Capacity Expansion: If domestic SMEs rapidly overcome the limitations of wire drawing technology through technology imports, global Q-fiber fabric production capacity may double to over 50 million meters by 2027, potentially narrowing the supply-demand gap and causing prices to fall by 15-20%.
2. Risk of Technological Iteration: Breakthroughs in dielectric properties and cost of new dielectric materials (such as polyimide film-modified materials) could replace existing Q-fiber fabric technologies.
3. Supply Chain Risk: Shortages of high-purity silica sand or geopolitical factors restricting imports will directly constrain Q-fiber fabric production capacity and may also affect customer certification and supply stability.
4. Risk of Lower-than-Expected Verification: Lower-than-expected Q-fiber fabric end-user verification progress in the first quarter (e.g., verification extended to the second quarter) will delay mass production, leading to lower-than-expected demand for the entire year.
5. Risk of Downstream Demand Fluctuations: Lower-than-expected capital expenditure in the AI server industry will impact demand for high-end electronic fabrics, suppressing industry prosperity.
