Lithography TAM is driven by leading-edge node transitions and fab capacity expansion. Each new node adds EUV intensity — more passes per wafer, higher ASP per tool — so revenue grows faster than unit volume. Base case assumes TSMC, Samsung, and Intel continue current capex trajectories, CHIPS Act fabs ramp on schedule, and High-NA EUV achieves volume production by 2028. Bear case assumes a semiconductor cycle downturn in 2027–2028 triggers capex freezes, delaying EUV orders 12–18 months. Bull case assumes AI infrastructure demands pull forward leading-edge capacity beyond current plans, High-NA adoption broadens to TSMC and Samsung simultaneously, and ASML scales EUV output above 70 units per year.
“ASML guided €30–35B revenue for 2025, driven by EUV demand from TSMC N2 ramp and High-NA EUV shipments to Intel and Samsung. Backlog exceeds €36B — more than a full year of revenue locked in. EUV systems now exceed €200M ASP per tool; High-NA tops €350M.”
“TSMC N2 mass production underway in Hsinchu. N2 requires ~14 EUV layers per wafer vs. ~10 at N3 — each node adds EUV intensity and compounds ASML revenue per wafer start. N2P and A16 (backside power delivery) are on roadmap through 2027.”
“DUV demand remains robust for trailing-edge nodes. Automotive, industrial, and AI chip packaging fabs are expanding 28nm–65nm capacity globally. DUV tools (ArF, KrF) are unaffected by EUV export controls and remain the workhorse for analog, power management, and mature-node logic.”
“High-NA EUV (ASML TWINSCAN EXE:5000) shipments to Intel 14A and Samsung commenced. High-NA eliminates multi-patterning at sub-2nm, reducing process steps and improving yield. ASP of ~€350M per tool creates a meaningful revenue step-up vs. standard EUV at ~€200M.”
“Netherlands, US, and Japan export controls restrict advanced DUV shipments to China. SMIC and Hua Hong face capacity ceilings at 7nm without new tool inflows. ASML China revenue declining as a share of total — manageable given non-China demand, but a meaningful market removed.”
“ASML EUV tool output constrained at ~60 units/year. Even with record demand, shipment cadence is gated by ASML's own supply chain — particularly Carl Zeiss optics and light source components. Demand is not the binding constraint; ASML's manufacturing capacity is.”
TSMC's N2 node is in mass production and A16 (1.6nm with backside power delivery) is on track for 2026. Each successive node adds EUV layers per wafer: N3 uses ~10, N2 uses ~14, A16 will use ~18+. This compounding EUV intensity means ASML's revenue per wafer start grows even if unit volume is flat. TSMC alone accounts for ~40% of ASML's total EUV revenue.
High-NA EUV (ASML TWINSCAN EXE:5000) is the enabling tool for sub-1.4nm patterning without extreme multi-patterning. At ~€350M per tool vs. ~€200M for standard EUV, each High-NA unit is ASML's highest-ASP product ever. Intel 14A is the first production customer; TSMC A14 and Samsung are expected to follow. If two or more leading-edge fabs commit simultaneously, ASML enters a new revenue S-curve.
TSMC Arizona (N4/N3 → N2), Intel Ohio and Arizona (18A/14A), Samsung Taylor (4nm/2nm), and Micron New York (DRAM) are all under construction with CHIPS Act subsidies. Each fab requires 20–60 EUV and DUV tools over a multi-year ramp. The US reshoring wave is the largest geographic expansion of leading-edge capacity since TSMC scaled in Taiwan — and ASML is the sole supplier for the tools that matter most.
Export controls blocking advanced ASML DUV shipments to China are pushing Chinese fabs toward domestic alternatives (SMEE) and exhausting approved inventories. Near-term: SMIC and Hua Hong face capacity ceilings at 7nm. Longer-term risk: if SMEE achieves 28nm immersion DUV by 2029, it removes a meaningful trailing-edge revenue opportunity from both ASML and Nikon.
Canon's FPA-1200NZ2C nanoimprint tool offers mechanical imprint patterning at a fraction of EUV cost (~$10M vs. €200M+). Canon has secured trial orders from memory makers. If NIL achieves sufficient resolution and throughput for logic or DRAM production, it would be the first credible alternative to ASML's EUV monopoly in decades. Most analysts view this as a 2030+ risk at earliest.