CITIC Securities: The proactive replenishment cycle is approaching; the fundamentals of specialty fabrics will accelerate their surpassing.

CITIC Securities Research Report states that by 2026, the demand for AI specialty fabrics will accelerate, and the supply-demand gap for Low DK Gen 2 and Low CTE is expected to expand to about 20%. At the same time, with the release of active inventory replenishment demand, the price elasticity of specialty fabrics is likely to increase. Our calculations show that specialty fabrics account for only 0.1% of server costs, and downstream sensitivity to price increases is relatively low. It is expected that by 2026, the prices of Low DK Gen 2 and Low CTE could double or more.

Full Text Below

Basic Materials | Active Inventory Replenishment Cycle Approaching, Fundamentals of Specialty Fabrics to Accelerate

By 2026, the demand for AI specialty fabrics will accelerate, and the supply-demand gap for Low DK Gen 2 and Low CTE is expected to expand to about 20%. Along with the release of active inventory replenishment demand, the price elasticity of specialty fabrics is likely to increase. Our estimates show that specialty fabrics account for only 0.1% of server costs, and downstream sensitivity to price increases is relatively low. It is expected that by 2026, the prices of Low DK Gen 2 and Low CTE could double or more.

▍ Accelerated AI demand in 2026.

North American CSP capital expenditure guidance continues to be revised upward, with a relatively optimistic outlook for 2026. We forecast global AI server spending to grow by 99% year-over-year in 2026. Based on expectations for Nvidia, ASICs, and switch shipments, we estimate that demand for specialty fabrics will reach approximately 215 million meters in 2026, with an incremental demand of about 120 million meters. In 2027, the incremental demand is expected to be around 110 million meters, with demand growth accelerating. Additionally, we expect Gen 2 fabrics to gradually become mainstream products, with their share of demand rising to about 35%/45% in 2026/27.

▍ Active inventory replenishment cycle approaching, price elasticity expected to increase.

Reviewing the last electronic fabric price increase cycle, demand growth was about 10%. According to Zhuochuang Information, an additional 130,000 tons of capacity was added, representing about a 17% increase. We believe that the extra 7% demand was absorbed by downstream active inventory replenishment, which also amplified price elasticity. Looking ahead, the 2025 downstream capital expenditure growth rate significantly exceeds that of 2020, and the demand driven by AI capital expenditure is expected to be larger and longer-lasting than the previous cycle. In 2026, AI downstream capacity deployment is likely to drive upstream inventory replenishment for specialty fabrics, further increasing their price elasticity.

▍ Prices of Low DK Gen 2 and Low CTE are expected to double or more.

We expect that before 2027, the high-end capacity supply-demand imbalance in AI PCBs will be clear, and high industry profit margins are likely to be maintained. Our calculations show that specialty fabrics account for only 0.1%–0.2% of end-server costs, so price increases in specialty fabrics will have a very limited impact on end customers’ profitability. We estimate that the supply-demand gap for Low DK Gen 2 and Low CTE could both exceed 20%. Under tight supply and demand conditions, combined with active inventory replenishment and good downstream profitability, prices for Low DK Gen 2 and Low CTE are expected to double or more in 2026.

▍ Similar to the quartz crystal years, the price ceiling for specialty fabrics may be higher than that of high-purity quartz sand.

High-purity quartz sand prices increased by 300% in the first half of 2023. Our estimates show that the supply-demand gap at that time was about 15% (including inventory replenishment). During the initial price rise of high-purity quartz sand, it accounted for only 1% of downstream silicon wafer costs. Currently, specialty fabrics account for only 0.1% of AI downstream costs, but AI downstream profitability is significantly higher than silicon wafers. Therefore, AI downstream companies are more accepting of price increases for specialty fabrics. Comparing the current situation to the high-purity sand price surge, both high-purity sand and Low DK Gen 2/Low CTE fabrics have supply-demand gaps exceeding 15%. High-purity sand benefits from a more favorable competitive landscape, while Low DK Gen 2 and Low CTE benefit from better downstream profitability and higher cost-end price acceptance. In comparison, we believe that the price ceiling for Low DK Gen 2 and Low CTE fabrics in 2026 could theoretically be higher than that of high-purity quartz sand.

▍ Risks:

Macroeconomic volatility, rising raw material and fuel costs, irrational capacity expansion, overseas demand recession, lower-than-expected demand and market expansion for high-end products, increasing industry competition.

▍ Investment Strategy.

We expect that in 2026, demand for specialty fabrics will reach about 215 million meters, with an incremental demand of approximately 120 million meters. In 2027, the incremental demand is expected to be around 110 million meters, with demand growth accelerating. The accelerated deployment of downstream AI capacity is likely to drive upstream inventory replenishment for specialty fabrics, further amplifying their price elasticity. Considering our estimate that specialty fabrics account for only about 0.1% of server costs, we believe that the prices of Low DK Gen 2 and Low CTE could both double or more in 2026.

(Source: First Financial)