Report Interpretation

Overview

This conference-takeaway report presents Bernstein's constructive view of ASML after meetings with its investor-relations team. The central conclusion is that Low-NA EUV output of 110 systems in 2028 reflects current demand and productivity assumptions rather than a hard supply limit, while strong order commitments, AI-related semiconductor demand and rising lithography intensity underpin Bernstein's Outperform view.

Core views

ASML reiterated a Low-NA EUV capacity roadmap of about 65 systems in 2026, 85 in 2027 and 110 in 2028. Management emphasized that these levels are derived from its current manufacturing footprint, sustainable-demand assumptions and expected productivity gains in next-generation systems, rather than a physical ceiling. Bernstein therefore sees scope for output above 110 units in 2028 if customer demand proves durable, especially as AI-related investment expands. Supply-chain coordination is part of the capacity case: Zeiss has agreed to raise capacity. For 2027, ASML expects roughly 30% growth in system shipments and about 45% growth in effective wafer capacity, aided by platform productivity; 2028 unit growth could remain near 30%, although productivity improvement is expected to be somewhat lower. The new Brainport campus is under construction but is not expected to add meaningful output until 2029; Veldhoven remains the EUV manufacturing and R&D hub, while Brainport will primarily support mature and DUV tools. Order visibility supports management's willingness to consider further expansion. The report says 2026 revenue is effectively covered by orders, 2027 EUV capacity is largely booked, and DUV capacity is close to fully allocated. ASML has also received 2028 orders, including customer down-payments of 50%, which Bernstein views as a stronger sign of commitment than in prior cycles. Management sees no current evidence of double-booking, but is monitoring customer requirements and shipment schedules to protect backlog quality. The report identifies pricing as a later-cycle opportunity rather than a near-term backlog driver. ASML says its value-based pricing reflects the throughput, overlay and accuracy delivered by newer tools, and seeks a larger share of the value customers create from advanced semiconductor manufacturing. Pricing gains are expected primarily in 2027-28 and particularly 2028; contractual terms mean they will not affect the existing backlog. Outcomes will still depend on customer negotiations and order volumes. ASML applies the same discipline to High-NA, arguing that lowering prices to speed adoption could weaken future pricing discipline. High-NA adoption is described as increasingly certain but differentiated by customer timing. ASML expects Samsung to adopt High-NA for DRAM around 2028, while TSMC may defer to about 2030 because it can extract strong utilization from Low-NA. Intel is already progressing with deployment. Management argues that customers will eventually need High-NA because continued scaling with Low-NA alone would materially impair yields; High-NA replaces complex multi-patterning with single-exposure lithography and can reduce manufacturing costs. For ASML, the economic difference between High-NA and optimized Low-NA is limited, and High-NA may initially be slightly dilutive, though profitability should improve materially as volumes rise. ASML remains constructive on lithography intensity in both logic and memory. It cites the difficulty of simultaneously changing transistor architectures and shrinking nodes, contrasting TSMC's temporary pause in scaling during the FinFET-to-GAA transition with Samsung's more challenging combined migration. In memory, management rejects the idea that the move from 6F² to 4F² will reduce lithography intensity: it expects additional EUV layers in critical DRAM steps and continued replacement of DUV multi-patterning with EUV single exposure. Memory could represent roughly 50% of future EUV demand and potentially sustain that share because AI-related memory production requires substantial wafer capacity. Productivity is another central demand and value driver. ASML, TSMC, Intel and Samsung are evaluating a potential move to 12-inch masks for High-NA. ASML estimates that larger masks could raise productivity by 40% through more efficient die layouts and alleviate some stitching constraints, though reticle-stage modifications remain technically difficult. Across wider platform improvements, management sees a path from roughly 330 wafers per hour on current Low-NA systems to 400-500 wafers per hour in future systems. The nearer-term F-system transition is expected to make F systems the dominant Low-NA mix by 2027; the F platform retains the 3800 optics architecture but uses a revised laser configuration to reach about 260 wafers per hour. The report links AI infrastructure to broader semiconductor demand. Management believes AI demand is spreading beyond the initial beneficiaries, TSMC and Hynix, to Intel, Samsung and Rapidus. It argues that enterprise AI, rather than consumer AI alone, is the key justification for the current infrastructure investment cycle, with physical-AI applications such as robotics providing a further potential source of demand. ASML also sees internal AI opportunities through its relationship with Mistral AI, including use of large internal data sets, improved R&D productivity and manufacturing optimization; successful adoption could reduce R&D intensity over time and improve output efficiency. On China, ASML warns that tighter export controls could accelerate investment in domestic Chinese lithography capabilities. Management nonetheless believes a substantial gap remains between Chinese prototype announcements and high-volume manufacturing, estimating Chinese EUV capability to be more than a decade behind industry leaders. It remains confident in its competitive position, citing technology leadership and strategic supplier relationships, notably Zeiss, while expressing caution that further restrictions, including the MATCH Act, could intensify domestic Chinese R&D. Finally, ASML highlights an expanding installed-base revenue stream. Roughly 95% of all systems ever shipped remain operational, creating a growing opportunity for services, software upgrades and productivity enhancements. EUV pay-per-wafer service revenue is estimated at around 5% of system ASP annually; High-NA is expected to have similar economics, while leading immersion systems generate recurring revenue of roughly 1-2% of ASP. Management sees these revenues as structurally growing as the installed base expands, although upgrades remain cyclically variable.

Analysis framework

Bernstein summarizes management discussions from its Semiconductor Day Conference, then tests the investment case through capacity and supplier readiness, customer commitments, pricing and product-roadmap economics, lithography intensity, AI-driven demand, competitive and export-control dynamics, and installed-base services. Its valuation derives a target price by applying a P/E multiple to its EPS estimate.

Methodology notes

• Valuation methodsP/E and PEG Valuation

  P/E multiple valuation

  Bernstein applies a 40x P/E multiple to its Q5-8 EPS estimate to derive a rounded €2,500 target price for ASML.

• Industry AnalysisSupply-demand framework

  Semiconductor equipment supply-demand analysis

  The report assesses EUV capacity against booked orders, customer deposits, supplier expansion, AI-related demand and expected tool productivity to judge whether output can grow beyond the stated 2028 plan.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Lithography value-chain transmission

  The report connects ASML tool throughput and lithography capability with customer manufacturing costs, yields, memory demand, supplier capacity and semiconductor scaling requirements.

Asset mapping & comparison

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

• ASML Holding NV (ASML.NA)

  Primary covered company; expected to benefit from strong EUV and DUV order visibility, AI-related semiconductor demand, expandable EUV capacity, productivity gains and service revenue growth.

  Strengths

  Technology leadership, booked capacity through 2027, 2028 customer down-payments, strategic supplier support from Zeiss, and a large installed base.

  Weaknesses

  High-NA could be slightly dilutive relative to optimized Low-NA tools before volumes scale.

  Comparison

  TSMC is expected to delay High-NA to roughly 2030 because of strong Low-NA utilization, while Samsung is expected around 2028 and Intel is already deploying High-NA.

  Risks

  Margins below expectations from EUV commercialization costs, China over-stocking, weaker WFE demand, slower technology migration, long-term competing technologies and tighter China export controls.

Key data

• ASML.NA closing price€1,387.80Close date 14 Sep 2026.
• ASML.NA price target€2,500.00Bernstein target price, implying 80% upside from the stated close.
• Low-NA EUV capacity roadmap65 systems in 2026; 85 in 2027; 110 in 2028Management says these are demand-led assumptions, not hard physical supply caps.
• 2027 system-shipment growthApproximately 30%Expected annual growth in system shipments.
• 2027 effective wafer-capacity growthApproximately 45%Supported by next-generation platform productivity improvements.
• Future EUV demand from memoryRoughly 50%ASML believes memory could account for and sustain about half of future EUV demand.
• Future system throughput400-500 wafers per hourManagement's pathway from roughly 330 wafers per hour on current Low-NA systems.
• EUV service revenueAround 5% of system ASP annuallyEstimated pay-per-wafer recurring service revenue.
• Revenue€32,667m in F25A; €43,973m in F26E; €55,915m in F27EBernstein financial estimates; stated CAGR is 30.8%.
• EBIT€11,302m in F25A; €17,737m in F26E; €24,280m in F27EBernstein financial estimates; stated CAGR is 46.6%.

Impact & implications

Bernstein's investment case rests on the possibility that durable customer demand and AI-related investment allow ASML to expand beyond its announced 2028 Low-NA EUV plan. The report also sees later-period pricing, improving platform productivity, increasing lithography intensity and a growing installed-base service stream as supportive factors, while High-NA timing affects product mix more than the overall strategic need for advanced lithography.

Risks

• Margins could fall below expectations because commercializing EUV and advancing it to new generations may cost more than expected.
• China over-stocking could weaken demand.
• The wafer-fab-equipment market could be weaker than expected.
• Technology migration could be slower than expected, and other technologies could become a long-term threat.
• Further tightening of export controls affecting China customers is a downside risk.

What to watch

• Whether durable customer demand permits Low-NA EUV capacity to rise beyond 110 systems in 2028.
• The quality of the backlog, including shipment schedules and any evidence of double-booking.
• High-NA adoption timing at Samsung, TSMC and Intel, and profitability improvement as High-NA volumes increase.
• Customer pricing negotiations and ordering volumes in 2027-28.
• AI infrastructure demand, particularly enterprise AI and AI-related memory requirements.
• The effect of export-control developments, including the MATCH Act, on China demand and domestic Chinese lithography investment.