Proliferating AI ASIC Designs and EUV Technology Upgrades Jointly Boost Mask Blank Demand, Benefiting HOYA and Lasertec

Overview

The report examines how the proliferation of AI ASIC designs and upgrades to advanced process nodes are driving growth in Japan's EUV mask blank industry and analyzes how HOYA and Lasertec stand to benefit. Bernstein believes that increases in the number of chip designs and EUV layers will expand mask blank consumption, while phase-shift masks and High-NA EUV will raise unit prices and technological barriers. HOYA benefits directly from its market share and next-generation technology advantages, while Lasertec benefits from stricter mask inspection requirements.

Core views

The report first breaks down EUV mask blank growth into two primary drivers: volume and unit price. As cloud service providers and AI companies develop their own GPUs, CPUs, and various AI accelerators, each chip design requires multiple sets of expensive EUV masks and mask blanks. Bernstein expects the number of chip designs using advanced logic nodes of 7nm and below to grow at a 21% CAGR from 2025 to 2028. TSMC's tape-out data corroborates the expansion of the design ecosystem: in December 2025, tape-out volume for its N2 node at a comparable development stage was approximately 2 times that of N3, rising to approximately 4 times by June 2026. Meanwhile, although TSMC did not increase the number of EUV layers when transitioning from 3nm to 2nm because of the shift to GAA architecture, EUV layer counts at advanced nodes are generally still increasing, from slightly above twenty layers currently to approximately 30 layers at A10 (1nm). The combination of more designs and more layers required per design is expected to drive EUV mask blank shipments from 17,500 units in 2025 to 30,000 units in 2028, representing a 19% CAGR. EUV mask blanks have already surpassed DUV products and account for approximately two-thirds of the total mask blank market. The report expects the EUV mask blank market to expand at a 25% CAGR from 2025 to 2028. Combining EUV and DUV, the mask blank market is projected to grow from US$1.33 billion in 2025 to US$2.23 billion in 2028, representing a 19% CAGR. Bernstein believes structural volume growth will continue beyond 2028, with the adoption of High-NA masks becoming a new growth driver. Unit price growth is being driven by the increasing complexity of mask structures. The average selling price of a single EUV mask blank is close to US$500,000, approximately 100 times the US$500 selling price of a DUV mask blank. EUV lithography uses reflective optics, requiring a mask blank with approximately 50 coating layers that form a multilayer reflector. The interlayer period is approximately 7 nanometers, and constructive interference is generated based on Bragg's law. Complex coatings, extremely stringent defect control, and manufacturing precision create high barriers to entry. As linewidths shrink, interference between adjacent reflected light beams reduces pattern contrast. Consequently, the industry is adopting phase-shift masks beginning at 2nm, using an additional phase-shift layer to change the phase of reflected light by 180 degrees and thereby improve contrast and pattern fidelity. The 2nm node may require approximately 5 phase-shift layers, rising to approximately 10 layers at 1.4nm. The average selling price of phase-shift masks is expected to be 20%—30% higher than that of conventional binary EUV masks. Their adoption has already begun with the start of 2nm mass production, providing near-term incremental growth for suppliers. The detailed report projects that upgrades to phase-shift masks and the adoption of High-NA will drive a high-single-digit CAGR in EUV mask blank average selling prices from 2025 to 2028. The next stage is High-NA EUV, whose numerical aperture increases from the current 0.33 to 0.55 and is expected to be used for DRAM and leading-edge logic nodes between 2027 and 2029. The report states that its resolution can improve by approximately 70%, enabling finer feature sizes with fewer multiple-patterning steps. For mask suppliers, High-NA requires new absorber-layer materials and designs, stricter defect control, greater mask precision, and more complex multilayer deposition processes. Consequently, unit prices are expected to exceed those of phase-shift masks. Over the longer term, anamorphic illumination High-NA masks may increase in size from the current 6×6 inches to 12×6 inches, with average selling prices potentially reaching more than twice those of existing mask blanks. HOYA is the most direct beneficiary. The EUV mask blank market is dominated by the HOYA-AGC duopoly, with HOYA holding more than 70% share. The report considers HOYA to hold the leading share at TSMC, Samsung, and SK hynix, while AGC historically held an advantage at Intel. Bernstein believes that as phase-shift masks and High-NA products account for a larger proportion of the market, the technology roadmap will increasingly favor HOYA. Its next-generation product capabilities may help the company gain further market share and increase its penetration at Intel. The report expects HOYA's EUV mask blank revenue to grow at a 26.2% CAGR from 2025 to 2028, faster than the overall market. In its updated company model, the EUV mask blank growth forecast for FY26/3—FY29/3E is raised from the previous 21% to 28%. The mask blank business has a greater impact on HOYA's profit than its revenue contribution would suggest. In FY26/3, the business accounted for 40% of HOYA's IT business revenue, 51% of IT business profit, and 33% of company-wide profit. Because both its growth rate and profit margin are higher than those of other businesses, Bernstein expects its share of IT business profit to rise to 55% in FY29/3E. Accordingly, the report raises HOYA's FY28/3E—FY29/3E total revenue forecasts by 4%—5% and operating profit forecasts by 4%—7%. DUV mask blanks remain a positive driver but are less important: constrained by capacity, HOYA relies mainly on mid-single-digit percentage unit price growth rather than meaningful volume growth. As approximately 65% of its mask blank revenue already comes from EUV-related products, overall growth will increasingly converge toward that of the faster-growing EUV market. Bernstein reiterates its Outperform rating on HOYA and continues to apply a 35 times Q5—Q8 forward P/E multiple, using its ¥1,013 forward EPS forecast to derive a target price raised from ¥34,800 to ¥35,500. Downside risks explicitly identified in the report include a slower EUV technology transition or changes in the competitive landscape, weaker cyclical HDD demand, and the impact of adverse exchange-rate movements on revenue. Lasertec's path to benefiting stems from inspection requirements becoming more stringent as mask complexity increases. The company is the sole supplier of the leading ABiCS mask blank inspection system and has launched its E120 and E320 platforms to support advanced nodes and future High-NA EUV demand. It also holds a sole-supplier position in actinic inspection of patterned EUV masks through ACTIS. However, the current revenue contribution from EUV mask blank inspection remains limited. Bernstein believes the more significant upside comes from the ACTIS and MATRiCS patterned mask inspection platforms and expects the company's relatively conservative earnings guidance to be gradually revised upward. The report reiterates its Outperform rating and ¥54,000 target price for Lasertec, based on a 40 times Q5—Q8 forward P/E multiple. Downside risks include slower adoption of actinic inspection at TSMC fabs, weaker-than-expected EUV lithography demand, uncertainty regarding competitors' market entry, and the effects of adverse exchange rates on revenue and margins.

Analysis framework

The report first estimates mask blank volumes based on the number of AI chip designs and EUV layer counts at advanced nodes. It then assesses changes in average selling prices based on the technology transition toward phase-shift masks, High-NA, and future anamorphic illumination masks, thereby producing forecasts for the EUV and overall mask blank markets. The report subsequently maps industry growth to HOYA's revenue and operating profit forecasts by incorporating market share, customer mix, technology roadmaps, and business profit contributions. For Lasertec, it analyzes equipment demand along the chain from increasing mask complexity to stricter inspection requirements. Finally, the report determines target prices for both companies using Q5—Q8 forward EPS and target P/E multiples.

Methodology notes

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Decomposition of mask blank sales volume and average selling price

  The report explains shipment growth using the number of advanced chip designs and EUV layer counts, and explains average selling price growth using the penetration of higher-value products such as phase-shift masks and High-NA, thereby deriving the market size.

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

  Demand transmission from AI chip design to mask blanks and inspection equipment

  The report traces the proliferation of chip designs by cloud service providers and AI companies through to tape-outs and EUV mask blank consumption, and then links increasing mask complexity to demand for Lasertec's inspection equipment.

• Industry/Sector Analysis FrameworkIndustry Concentration Analysis

  Analysis of the EUV mask blank duopoly and changes in market share

  The report compares HOYA and AGC in terms of market share, customer base, and next-generation technology capabilities to assess whether HOYA can continue gaining share during the transition to phase-shift masks and High-NA.

• Valuation MethodPE/PEG valuation

  Q5—Q8 forward P/E valuation

  HOYA's target price is based on a 35 times Q5—Q8 forward P/E multiple and ¥1,013 in forward EPS; Lasertec's target price is based on a 40 times Q5—Q8 forward P/E multiple.

Asset mapping & comparison

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

• HOYA (7741.JP)

  Expansion of the EUV mask blank market and upgrades to phase-shift masks and High-NA directly drive its revenue, profit, and potential market share growth.

  Strengths

  Holds more than 70% of the EUV mask blank market and has technological leadership in phase-shift masks and High-NA; the mask blank business has higher margins and growth than the company's other businesses.

  Weaknesses

  The DUV mask blank business is capacity-constrained, with growth relying mainly on unit prices; some of the company's businesses remain exposed to cyclical HDD demand.

  Comparison

  The EUV mask blank market is dominated by the HOYA-AGC duopoly. The report believes HOYA holds the leading share at TSMC, Samsung, and SK hynix, while AGC historically had a stronger position at Intel.

  Risks

  A slower-than-expected EUV technology transition, changes in the competitive landscape, weaker HDD demand, and adverse exchange-rate movements.

• Lasertec (6920.JP)

  Increasing complexity in EUV masks and mask blanks creates more stringent inspection requirements, supporting long-term demand for its inspection equipment.

  Strengths

  It is the sole supplier of the ABiCS mask blank inspection system and maintains a sole-supplier position in actinic patterned mask inspection through ACTIS; E120 and E320 can support advanced nodes and future High-NA demand.

  Weaknesses

  The current revenue contribution from EUV mask blank inspection remains limited, with the main upside depending on the ACTIS and MATRiCS patterned mask inspection platforms.

  Comparison

  The report emphasizes the company's sole-supplier positions in ABiCS and ACTIS but does not provide a quantitative peer comparison.

  Risks

  Slower adoption of actinic inspection by TSMC, weaker-than-expected EUV lithography demand, uncertainty regarding competitors entering the market, and the impact of adverse exchange rates on revenue and margins.

Key data

• EUV Mask Blank Market Growth25% CAGR from 2025 to 2028Jointly driven by the number of designs, EUV layer counts, and penetration of higher-value products
• Total Mask Blank Market SizeUS$1.33 billion in 2025; US$2.23 billion in 2028ECombined EUV and DUV, with an expected CAGR of 19%
• Number of Advanced-Node Chip Designs21% CAGR from 2025 to 2028Covering nodes of 7nm and below
• TSMC N2 Tape-Out VolumeApproximately 4 times that of N3 during the same periodApproximately 2 times in December 2025, rising to approximately 4 times by June 2026
• Advanced-Node EUV Layer CountIncreasing from slightly above twenty layers to approximately 30 layers at A10 (1nm)The increase in layer count raises the number of mask blanks required for each chip design
• EUV Mask Blank Shipments17,500 units in 2025; 30,000 units in 2028EExpected CAGR of 19%
• EUV Mask Blank Unit PriceClose to US$500,000 per unitApproximately 100 times the US$500 unit price of a DUV mask blank
• Phase-Shift Mask Unit Price Premium20%—30%Relative to conventional binary EUV masks
• EUV Mask Blank Average Selling Price GrowthHigh-single-digit percentage CAGR from 2025 to 2028EForecast in the detailed report, driven mainly by phase-shift mask and High-NA upgrades
• High-NA Parameters and Resolution0.55 NA versus the current 0.33 NA; approximately 70% improvement in resolutionExpected to support DRAM and leading-edge logic nodes between 2027 and 2029
• HOYA EUV Mask Blank Market ShareMore than 70%Forms a duopoly with AGC
• HOYA EUV Mask Blank Revenue Growth26.2% CAGR from 2025 to 2028EExpected to outpace the overall EUV mask blank market
• HOYA Updated Growth Forecast28% CAGR from FY26/3 to FY29/3EPrevious forecast was 21%
• HOYA Estimate RevisionsRevenue raised by 4%—5%; operating profit raised by 4%—7%Applicable to FY28/3E—FY29/3E
• HOYA Mask Blank Business Contribution40% of IT revenue, 51% of IT profit, and 33% of total profit in FY26/3Expected to rise to 55% of IT profit in FY29/3E
• HOYA Target Price¥35,500Previous target price ¥34,800; based on a 35 times Q5—Q8 P/E multiple and ¥1,013 in forward EPS
• Lasertec Target Price¥54,000Based on a 40 times Q5—Q8 forward P/E multiple

Impact & implications

The report argues that the proliferation of AI ASIC designs not only increases the number of advanced-node tape-outs but also amplifies mask demand for each design through higher EUV layer counts. Phase-shift masks and High-NA extend growth beyond pure volume expansion to higher unit prices and technological barriers. With more than 70% market share, strong next-generation technology capabilities, and a high profit contribution from mask blanks, HOYA is expected to outgrow the industry. Lasertec benefits from increasingly stringent defect-control and pattern-fidelity requirements, although its near-term earnings upside depends more heavily on the ACTIS and MATRiCS patterned mask inspection platforms.

Risks

• A slowdown in the EUV technology transition could weaken mask blank demand and alter HOYA's competitive landscape.
• Changes in cyclical HDD demand could create downside pressure for HOYA.
• Adverse exchange-rate movements could reduce HOYA's revenue and affect Lasertec's revenue and margins.
• Slower-than-expected adoption of actinic inspection at TSMC fabs could affect demand for Lasertec's equipment.
• Weaker-than-expected EUV lithography demand could reduce growth in Lasertec's related inspection platforms.
• Changes in the timing of competitors' market entry could affect Lasertec's competitive position.

What to watch

• Track the number of advanced-node chip designs and whether growth in TSMC N2 tape-outs relative to N3 can continue.
• Monitor progress in increasing EUV layer counts from slightly above twenty layers to approximately 30 layers at 1nm.
• Monitor the production ramp of 2nm phase-shift masks and the increase in phase-shift layer counts at the 1.4nm node.
• Monitor the adoption progress of High-NA EUV in DRAM and leading-edge logic nodes between 2027 and 2029.
• Track whether HOYA can leverage phase-shift mask and High-NA technologies to gain market share and increase its penetration at Intel.
• Monitor potential upward revisions to Lasertec's conservative guidance and changes in demand for the ACTIS and MATRiCS platforms.