High-end fiberglass product upgrades combined with AI server demand should accelerate market share gains for Taiwan CCL manufacturers

Overview

Starting with a meeting with Fulltech management, the report analyzes the revenue and profit impact of shifting fiberglass capacity from E-glass toward Low DK, Low DK2, and Low CTE, and extends its conclusions to Taiwan's CCL industry. Goldman Sachs believes AI servers and CCL specification upgrades are creating a supply shortage in high-end materials, benefiting Fulltech, Elite Material, and TUC, which possess high-end product capabilities; by contrast, demand for mid- to low-end CCL remains weak.

Core views

Fulltech plans to expand fiberglass capacity by more than 60% by 2029E and add fiberglass yarn capacity for Low DK, Low DK2, and Low CTE (T-glass), gradually reducing the share of E-glass. The company expects more than 50% of new fiberglass capacity through 2028E to come from its new Thailand plant. Half of the new Thailand capacity is planned for LEO applications, while the other half will be used for AI-related Low DK and Low DK2 products. If demand exceeds expectations, the company may further increase capacity allocation to Low DK and Low DK2. The Thailand expansion also addresses LEO customers' requirements for capacity outside China, and the currently planned capacity is already fully booked. Product mix upgrades are the key drivers of revenue and profit. In 1H26, E-glass accounted for approximately 25% of Fulltech's revenue, while Low DK, Low DK2, and Low CTE together accounted for approximately 55%. The company has observed strengthening Low DK2 orders and continues to convert Low DK capacity to Low DK2; by 2H26, Low DK2 could account for more than 50% of combined high-end fiberglass shipments comprising Low DK and Low DK2. Price differences are substantial: Low DK is priced at approximately 5 times E-glass; before E-glass prices more than doubled from 2025 onward, the price gap was approximately 10 times. Goldman Sachs expects Low DK2 to be priced at more than twice Low DK. In terms of gross margin, Low DK2 ranks highest, followed by Low DK, while E-glass remains the lowest even after substantial price increases. Goldman Sachs uses a scenario in which all capacity is converted to Low DK and Low DK2 to assess the mix effect: without assuming further price increases, revenue could more than double, while gross margin could rise from approximately 38% in 1H26 to 45%-50%. Lower output would not fully offset these gains because converting E-glass to Low DK reduces fiberglass output by approximately 50%, while converting Low DK to Low DK2 reduces output by 10%-20%; however, higher average selling prices and gross margins are expected to be sufficient to offset the shipment loss and continue driving revenue growth and margin expansion. Upgrade progress varies across plants. Approximately 90% of revenue at Fulltech's Plant 1 in Taiwan came from Low DK and Low DK2 in 1H26, with only approximately 10% coming from E-glass; the plant will continue shifting from Low DK toward stronger-demand Low DK2 in 2H26. Revenue at Plant 2 in mainland China came entirely from E-glass during the same period, but the plant plans to shift toward Low DK and Low DK2 in 2027, with these products potentially accounting for as much as 50% of the plant's total revenue. The company also plans to expand the plant's fiberglass capacity by more than 15% by 2029E, focusing on Low DK, Low DK2, and Low CTE. Low CTE represents another source of growth. Fulltech has already begun shipping Low CTE fiberglass yarn, with shipments continuing to increase, and is also qualifying Low CTE fiberglass. Customers have asked the company to double related capacity in 2H26 amid Elite Material's expansion of substrate CCL capacity and tight substrate material supply. As Low CTE offers better pricing and margins than existing products, Goldman Sachs expects it could become Fulltech's highest-gross-margin product after mass production ramps up. At the industry level, high-end CCL demand is driven primarily by AI servers, with new AI projects expected to ramp up in 2H26. M7+ grade CCL uses Low DK and Low DK2 fiberglass, so specification upgrades directly improve the product mix of fiberglass suppliers with the relevant capabilities. Goldman Sachs expects high-end CCL demand to grow at a 96% CAGR from 2025 to 2028E, versus a supply CAGR of only 22%. With high-end CCL supply constrained, manufacturers can pass the additional costs of key raw materials such as HVLP copper foil and Low DK2 fiberglass downstream and are also willing to pay premiums to secure supply, creating a more favorable pricing environment for high-end raw materials. The dynamics for mid- to low-end CCL are different. Prices in this segment rose even faster than those of high-end products in 1H26, but this was driven mainly by higher raw material costs and tighter low-end CCL supply rather than strong demand; Goldman Sachs expects mid- to low-end CCL demand to grow at a CAGR of only 1% from 2025 to 2028E. As suppliers shifted capacity toward Low DK, E-glass prices have risen by more than 80% year to date, resulting in a pronounced shortage. The report expects low-end CCL prices to continue rising in 3Q26, but the pace of price increases should slow in 4Q26: on the one hand, suppliers may shift some capacity back to E-glass, easing the shortage; on the other hand, consumer electronics, which account for more than 50% of low-end CCL demand, may experience weak seasonality in the fourth and first quarters. Persistently weak mid- to low-end demand could ultimately slow E-glass demand and weaken its pricing environment. For Elite Material, Goldman Sachs highlights its market shares of more than 70% in high-end HDI materials and more than 90% in SLP materials, as well as its 3-5+ years of experience in the high-speed switch and server CCL markets. The report expects its server market share to rise from less than 10% on the Purley platform to 15%-20% on Whitley and more than 20% on Eagle Stream; its switch market share also increased from less than 5% before 2019 to more than 30% in 2023 and 40% in 2024. Its AI customer base expanded from just one 400G customer in 3Q20 to becoming one of the major suppliers in 2023, and it is described as a major supplier to Nvidia, Google, and AWS AI server projects. Goldman Sachs believes these advantages support long-term revenue growth and gross and operating margin expansion, and maintains a Buy rating and a 12-month target price of NT$10,200; the target price applies a 27 times 2H27-1H28E P/E multiple, in line with the peak P/E multiple of AI component suppliers over the past three years. For Taiwan Union Technology Corp., the company focuses on M7+ high-speed CCL, with applications covering high-end switches above 100G and AI server materials, and has held more than 20% global market share over the past two years. The company expects its unit share in the server market to rise further on the Eagle Stream platform from 15% on the Whitley and Purley platforms and plans to launch T2A and T2C products for low-end and high-end customers. Growth in 400G and 800G switch shipments, new 800G switches planned for launch from 2025 to 2027, and high-end AI projects developed with CSP and enterprise customers are viewed as medium- to long-term revenue drivers. Goldman Sachs believes TUC's P/E valuation is low relative to its Taiwan CCL peers and maintains a Buy rating and a 12-month target price of NT$3,010; the target price is based on 22 times 2027E P/E, two standard deviations above the industry's average P/E over the past three years.

Analysis framework

The report first organizes Fulltech management's comments on capacity, product mix, plant conversion progress, and customer demand, and then estimates the economic impact of product upgrades based on the pricing, gross margins, and output changes of different fiberglass products. Goldman Sachs subsequently compares the demand, supply, and pricing drivers of high-end versus mid- to low-end CCL, traces upstream fiberglass changes through to CCL manufacturers, and finally determines ratings and target prices based on Elite Material's and TUC's market shares, customer projects, growth prospects, and forward P/E multiples.

Methodology notes

• Industry/Sector Analysis FrameworkSupply-demand framework

  Comparison of supply and demand growth for high-end and mid- to low-end CCL

  The report separately compares demand growth, supply expansion, and the sources of shortages for high-end and mid- to low-end CCL to explain the pricing power of different products. From 2025 to 2028E, high-end CCL is expected to record a demand CAGR of 96% and a supply CAGR of 22%, while the demand CAGR for mid- to low-end products is only 1%.

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Combined assessment of output losses and improvements in pricing and gross margin

  The report considers both the decline in output caused by capacity conversion and the higher pricing and gross margins of Low DK and Low DK2 to determine whether product mix improvement is sufficient to offset shipment losses.

• Industry/Sector Analysis FrameworkUpstream-Midstream-Downstream Value Chain Transmission

  Transmission of fiberglass raw material costs to CCL prices and manufacturers' profits

  The report tracks supply and price changes in raw materials such as Low DK2 fiberglass and HVLP copper foil and analyzes how high-end CCL manufacturers pass on costs through price increases, pay premiums to secure raw materials, and affect upstream suppliers through these actions.

• Valuation MethodPE/PEG valuation

  Forward P/E target-price method

  Elite Material's 12-month target price applies a 27 times 2H27-1H28E P/E multiple; TUC applies a 22 times 2027E P/E multiple, using the peak valuation of AI component suppliers and the industry's historical average valuation as respective benchmarks.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Fulltech (1815.TWO)

  As the subject of the management meeting, its product mix and capacity expansion plans are the main basis for the report's assessment of high-end CCL raw material trends in Taiwan; Goldman Sachs does not cover the stock.

  Strengths

  Low DK, Low DK2, and Low CTE offer significantly higher pricing and gross margins than E-glass, Thailand's LEO capacity is fully booked, and customers have requested an expansion of Low CTE capacity.

  Weaknesses

  Capacity upgrades reduce physical output, with conversion from E-glass to Low DK reducing output by approximately 50% and conversion from Low DK to Low DK2 reducing output by 10% to 20%.

  Comparison

  Its high-end product mix is superior to a model relying solely on E-glass, and upgrade progress at Taiwan Plant 1 is significantly faster than at mainland China Plant 2.

  Risks

  The report does not list company-specific downside risks for Fulltech.

• Elite Material (2383.TW)

  A key supplier of high-end HDI, SLP, and AI server CCL, expected to benefit from tight high-end CCL supply, customer expansion, and rising server market share.

  Strengths

  Its market shares in high-end HDI materials and SLP materials exceed 70% and 90%, respectively, and the report describes it as a major supplier to Nvidia, Google, and AWS AI server projects.

  Weaknesses

  High-end smartphone HDI materials face potential substitution by RCC.

  Comparison

  The report believes its 2026E P/E is below that of AI server component suppliers and, combined with its growth prospects, offers an attractive valuation.

  Risks

  RCC substitution of high-end smartphone HDI designs, trade tensions weakening smartphone and server shipments, and rising competition from mainland Chinese peers.

• Taiwan Union Technology Corp. (6274.TWO)

  An M7+ high-speed CCL supplier benefiting from growth in AI servers and 400G/800G switches, as well as rising market share on server platforms.

  Strengths

  Its high-end CCL market share has exceeded 20% over the past two years, and it is advancing high-end AI projects with CSP and enterprise customers.

  Weaknesses

  The investment thesis depends on continued market share gains in low-loss CCL and the scheduled ramp-up of new switches and AI projects.

  Comparison

  Goldman Sachs believes its P/E valuation is below that of its Taiwan CCL peers.

  Risks

  Slower-than-expected market share gains in low-loss CCL, trade tensions weakening global server and switch shipments, and rising competition from mainland Chinese peers.

Key data

• Fulltech fiberglass capacity expansion planIncrease by more than 60% by 2029EIncludes new capacity related to Low DK, Low DK2, and Low CTE
• Contribution from the new Thailand plantMore than 50% of new fiberglass capacity by 2028EApproximately 50% planned for LEO and approximately 50% for AI-related Low DK and Low DK2
• Fulltech 1H26 revenue mixE-glass approximately 25%; Low DK, Low DK2, and Low CTE combined approximately 55%Shows that high-end products have become the primary revenue source
• Low DK2 shipment shareCould exceed 50% of high-end fiberglass shipments in 2H26High-end fiberglass includes Low DK and Low DK2
• Product price differentialLow DK is approximately 5 times E-glass; Low DK2 is expected to exceed 2 times Low DKE-glass prices have more than doubled since 2025
• Full high-end conversion scenarioRevenue could more than double, with gross margin reaching 45% to 50%Compared with a gross margin of approximately 38% in 1H26 and assuming no further price increases
• Output impact of capacity conversionConverting E-glass to Low DK reduces output by approximately 50%; converting Low DK to Low DK2 reduces output by 10% to 20%The report believes higher pricing and gross margins are sufficient to offset the output loss
• Taiwan Plant 1 revenue mixApproximately 90% from Low DK and Low DK2 and approximately 10% from E-glass1H26 data
• Mainland China Plant 2 upgrade planLow DK and Low DK2 revenue share could reach as high as 50% in 2027All of the plant's revenue came from E-glass in 1H26, and capacity is planned to expand by more than 15% by 2029E
• High-end CCL demand and supply growthDemand CAGR of 96%; supply CAGR of 22%2025 to 2028E
• Mid- to low-end CCL demand growth1% CAGR2025 to 2028E
• E-glass price changeUp more than 80% year to dateThe shortage was caused by suppliers shifting capacity toward Low DK
• Elite Material market positionMore than 70% share in high-end HDI materials; more than 90% share in SLP materialsMarket shares in key subsegments as stated in the report
• Elite Material target price12-month NT$10,200Based on 27 times 2H27-1H28E forward P/E
• TUC market shareMore than 20% over the past two yearsGlobal market share in M7+ high-speed CCL
• TUC target price12-month NT$3,010Based on 22 times 2027E forward P/E

Impact & implications

The report believes that shifting fiberglass capacity from E-glass toward Low DK, Low DK2, and Low CTE will offset lower output through higher pricing and gross margins, improving Fulltech's revenue and earnings mix. AI servers and high-speed switches are driving high-end CCL demand far faster than supply, thereby strengthening the bargaining power of high-end raw material suppliers and supporting continued market share gains for Elite Material and TUC; however, price increases for mid- to low-end CCL lack support from strong end demand, and the subsequent pricing environment could weaken as the E-glass shortage eases.

Risks

• RCC could fully replace HDI designs in high-end smartphones, affecting Elite Material's related business.
• Escalating trade tensions could cause global smartphone, server, and switch shipments to fall below expectations.
• Rising competition from mainland Chinese peers could compress the market shares or profitability of Elite Material and TUC.
• TUC's market share gains in low-loss CCL could be slower than expected.

What to watch

• Monitor whether Fulltech can increase Low DK2 to more than 50% of high-end fiberglass shipments in 2H26.
• Track Low CTE fiberglass qualification, fiberglass yarn shipment growth, and progress toward the customer-requested doubling of capacity in 2H26.
• Monitor construction progress at Fulltech's new Thailand plant through 2028E and the actual allocation of capacity between LEO and AI-related products.
• Monitor the progress of mainland China Plant 2's conversion from E-glass to Low DK and Low DK2 in 2027.
• Track whether the pace of low-end CCL price increases slows as expected in 4Q26 after further increases in 3Q26.
• Monitor the ramp-up of AI projects in 2H26 and whether the growth gap between high-end CCL demand and supply persists.
• Track changes in Elite Material's and TUC's market shares in servers, 400G/800G switches, and high-end AI projects.