The Zhitong Finance App learned that CITIC Construction Investment released a research report saying that it is optimistic that LDK2 will become the variety with the greatest potential application space in the future in special distribution. LDK2 distribution combines application continuity and demand growth in platform iterations. Rubin and subsequent generations of LDK2 single GPU consumption can be increased to about 2.8 times or more than Blackwell's. Furthermore, I am optimistic that the LDK2 generation application penetration rate will increase in ASIC servers, and the 1.6T switch volume will drive demand growth. According to estimates, the LDK2 distribution supply and demand gap will continue to expand in 2026-2028, and the gap will exceed 90 million meters in 2028. The bank expects batch application of 2027H2 domestic looms in the production of ordinary cloth. I am optimistic that the new technology will be implemented first.
CITIC Construction Investment's main views are as follows:
The bank is optimistic that LDK2 cloth will become the variety with the largest potential application space in the future among special electronic fabrics
By dismantling the PCBBOM of various generations of Nvidia AI servers, the bank discovered that LDK2 distribution is the main type of deployment in the Nvidia platform iteration process, and has both continuity of application and demand growth across generations. Rubin and subsequent generations of LDK2 single GPU consumption can be increased to about 2.8 times or more than Blackwell's. In addition, the bank is optimistic that the application penetration rate of LDK2 distribution in ASIC servers will continue to increase, and the future volume of 1.6T switches will also drive the growth in demand for LDK2 proxy distribution.
LDK2's supply and demand gap continues to expand in 2026-2028, and the 2028 gap exceeds 90 million meters
According to estimates: (1) LDK1 distribution: 2026-2028 supply and demand gap rates are -2.1%/-18.1% (supply and demand gap ratio is positive, negative represents easing, same below), 2026-2027 maintains a tight balance between supply and demand, and tends to ease in 2028; (2) LDK2 distribution: 2026-2028 supply and demand gaps are 9.8%/25.9%/40.7%, respectively. The 2026-2028 supply and demand gap continues to expand, and the 2028 gap exceeds 0.9 billion meters (3 qb); :2026- Demand in 2028 was 31.6/531.9/32.018,000 meters, respectively, and 2028 ushered in demand volume, and attention should be paid to its technical route competition with the unwoven solution; (4) T cloth: 2026-2028 supply and demand gaps are 8.0%/-1.3%/5.0%, respectively, and 2026-2028 overall supply and demand are in a tight balance; (5) Ordinary electronic cloth: 2026-2028 supply and demand gaps are -0.3%/-6.3%/-7.0%, respectively. Supply and demand will be relaxed from 2027 to 2028 Releasing+imitation cloth shocks will put potential pressure on future prices.
It is expected that 2027H2 domestic looms will be used in batches in the production of ordinary electronic cloth
With overseas looms such as Toyota and Tsudakoma giving priority to the production of special electronic fabrics, the 2027-2028 loom supply bottleneck can be effectively mitigated by domestic manufacturers such as Shandong Rifa and Zhejiang Wanli. According to estimates, in 2026-2028, the increase in demand for looms driven by demand for special electronic cloth was 1,185/2,222/2,378 units, respectively. When the effective supply of overseas looms was 1,275/1,775/2,075 units, respectively, the increase in demand for looms to meet the conversion of special electronic cloth plus demand for ordinary electronic cloth was 639/2,094/2,501 units, respectively. The bank expects that after 2027H2, looms produced by domestic loom manufacturers can meet the performance requirements for the production of ordinary electronic cloth and begin batch supply. Assuming that domestic loom manufacturers effectively supply 3,600/7,200 units in 2027-2028, the above increase in demand can be met.
New technology for special electronic fabrics is evolving at an accelerated pace, and Wufu is expected to be the first to be launched
As the performance requirements of high-end PCBs such as low dielectric loss and dimensional stability continue to improve, third-generation cloth, organic cloth, and non-cloth have become the three main new technology directions. Third-generation cloth is still in the early stages of industrialization, and high-end PCB applications of organic cloth have yet to be verified; in contrast, the downstream application scenarios of unwoven fabrics are more clear (NV cabinet+high-speed optical module), and are expected to be implemented first.
Investment advice
On the upstream material side, it is recommended to focus on [Sinoma Technology], [Honghe Technology], [China Jushi], and [Wright Optoelectronics]; for high-performance resins, it is recommended to focus on [Shengquan Group] [Dongcai Technology]; on the CCL/PP side, it is recommended to focus on [Huazheng New Materials]; on the PCB manufacturing side, it is recommended to focus on [Shenghong Technology], [Shennan Circuit] and [Jingwang Electronics].