Sinolink: The second half of the year will see a dual benefit of new product launches and profit recovery in the PCB sector.

Sinolink has released a research report stating that the PCB industry is expected to experience a dual benefit of new product launches and profit recovery in the second half of the year. The iterative development of orthogonal backplane technology is likely to support the medium to long-term growth of the PCB industry, while CoWoP transfers the value of the substrate to PCB. Overall, the fundamentals of upstream materials in the PCB industrial chain in the first half of this year significantly outperformed those of downstream PCBs. However, the firm believes that the PCB segment will see dual marginal changes in new product launches and profit recovery in the second half of the year, with a strong certainty of accelerating fundamental growth. The current configuration value of the sector is relatively high, and it is recommended to pay attention to it. Sinolink's main points are as follows: Upstream materials accelerated in the first half of the year The acceleration of upstream material performance started in Q1 2026, with significant differences in growth rates for revenue/net profit attributable to parent company/non-recurring net profit between Q1 2026/Q4 2025 and Q2 2026/Q1 2026. The quarter-on-quarter growth rate changes for upstream materials between Q2 2026 and Q1 2026 narrowed significantly; the peak change in the quarter-on-quarter growth rate for the CCL segment occurred in Q2 2026. According to the data, the quarter-on-quarter growth rate changes for revenue/net profit attributable to the parent company/non-recurring net profit in Q1 2026/Q4 2025 were -3 pct/-428 pct/-433 pct, while for Q2 2026/Q1 2026 they were 37 pct/188 pct/157 pct, indicating acceleration starting in Q1 2026. The PCB segment in the first half of 2026 did not yet show a trend of accelerated initiation, with the quarter-on-quarter growth rate changes for revenue/net profit attributable to the parent company/non-recurring net profit in Q1 2026/Q4 2025 being 10 pct/163 pct/57 pct, and in Q2 2026/Q1 2026 being 31 pct/4 pct/18 pct, maintaining overall stability. The reasons for this phenomenon are: 1) No large-scale product demand for high-end AI products in the first half of the year. High-end AI products such as NVIDIA's Rubin, Amazon's new generation of Trainium 3 products, and Google's TPU V8 series saw shipments in the second half of 2026, resulting in a supply vacuum in the first half. This factor had a greater impact on A-share PCBs than on midstream and upstream. 2) AI demand diverted to lower-end products. Although the pushing force for AI new products was insufficient in the first half of this year, customers began to notify limited stockpiling at the end of March and comprehensive stockpiling in May. Additionally, due to most manufacturers in the industrial chain being optimistic about AI's forward layout and having sufficient motivation to shift capacities for AI product R&D testing, upstream material capacities for low-end PCB products became severely insufficient, leading to frequent price increases. The second half of the year is expected to usher in dual benefits of new product launches and profit recovery for PCBs The firm believes that starting from the third quarter, the performance slope of PCBs is likely to accelerate, making PCBs the most impressive sector. The core reasons are: 1) High-end AI new products have started to ship. According to industry chain tracking, the firm has observed that some high-end AI new products achieved shipments by the end of July, with the latest part entering mass production by mid-August. This strong demand is expected to continue until the end of this year, and PCBs, being the largest segment exposed to AI, will significantly benefit. 2) PCB prices are beginning to be transmitted. After price increases in midstream and upstream reached the PCB sector in the first half of the year, PCBs have not fully passed on these price increases to customers. However, the firm has noted that this situation is actively improving, and from an industrial chain tracking perspective, PCB profit levels are expected to further increase in the second half of the year, pushing the acceleration slope upwards. Data shows that monthly revenue growth rates of Taiwanese PCB manufacturers have already shown significant changes, with July 2026 seeing a year-on-year growth rate of 39%, achieving the highest growth slope this year, and a month-on-month growth rate of 13%, indicating that a turning point has emerged. Technological innovation continues, supporting growth in the PCB industry Even as technological innovation continues, the growth in value within the PCB industry can ensure the sustainability of its performance, regardless of fluctuations in AI. Thus, whether technical iterations can continue will become an important judgment anchor point. Ongoing technological iterations in AI PCB are still visible: 1) PCB orthogonal backplane. Currently, under NVIDIA's Rack architecture, the interconnection between the Compute Tray and NVLink Switch Tray is completed via copper cables, but issues such as poor heat dissipation and extreme assembly difficulty persist. The industry is also developing new solutions to achieve better mass production results. The firm's observations indicate that the previously used PCB orthogonal backplane in communication fields is one of the important directions to solve this issue. Once this solution is adopted, the value of PCBs is expected to further upgrade, thus supporting the long-term growth of the PCB industry. 2) CoWoP shifts the value from substrates to PCBs. The CoWoP design eliminates the packaging substrate stage and connects directly with PCBs, enhancing the obscurity of PCBs. While it is uncertain when these products will become mainstream, this shows that overseas terminal manufacturers still highly value PCB technology in their future design solutions, serving as a long-term support for the evolution of PCB technology. Risk warning AI capital expenditures may fall short of expectations; changes brought by model innovations such as DeepSeek; AI technology development may not meet expectations; intensifying competition.