TSMC and ASML Advance 12-Inch Photomask Development, Plan Pilot Production Line by 2031 to Unlock High-NA EUV Full-Field Exposure
nashnova research
TSMC and ASML are jointly developing a 12-inch photomask format, targeting a pilot line by 2031 and production tools by 2033, to restore High-NA EUV's usable exposure area from 429 mm² to 858 mm² — determining whether next-generation AI GPUs can be patterned in a single shot and reshaping the cost structure of leading-edge chips.
Where exactly is High-NA EUV stuck today?
High-NA EUV — the next-generation extreme-ultraviolet lithography system, higher resolution but a smaller field — raises the numerical aperture from 0.33 to 0.55, improving single-exposure resolution from ~13 nm to 8 nm.
The trade-off: an asymmetric reduction design (4× in X, 8× in Y) cuts the exposure field in half. With today's 6-inch mask, usable area is only about 429 mm².
This means → mainstream AI GPU dies run 750–850 mm². One shot cannot cover them; multi-exposure stitching is required, adding time and cost.
How does the 12-inch mask fix this?
TSMC and ASML propose a 6 × 12-inch mask format that restores full-field exposure to 858 mm² — enough to pattern today's largest chips in a single shot.
ASML estimates full-field single exposure improves throughput by roughly 40%.
In plain terms = the mask is finally big enough — no stitching, no yield risk at stitch boundaries, and fewer mask sets per critical layer.
Why can't they just scale up the mask?
Going from 6-inch to 12-inch is not a simple enlargement: the mask blank, fabrication process, inspection tools, pellicle — a thin membrane that shields the mask from particles — and mask pod all need redesign.
This means → the entire supply chain must upgrade in sync; a bottleneck at any single link delays the whole program.
Timeline: 2031 for a pilot line, 2033 for production-grade tools — less than six years for the industry to get ready.
How concentrated is the supply chain?
ASML is the sole EUV scanner supplier; Zeiss SMT provides the core optics; Cymer (an ASML subsidiary) supplies the drive laser — three firms lock down the equipment side.
Mask blanks are dominated by HOYA, AGC, and Shin-Etsu Chemical; on the inspection side, Lasertec holds an especially strong position in actinic (at-wavelength) EUV inspection.
This reflects a broader pattern: Japanese suppliers hold over 80% of the EUV photoresist and chemical-materials market, and pellicle supply depends heavily on Mitsui Chemicals — the 12-inch mask timeline ultimately hinges on whether this Japan-dominated materials chain can keep pace.
When do the major fabs plan to use it?
Intel targets the 18A node; TSMC aims at A10; Samsung Foundry is set for its 1 nm process.
On the memory side: Samsung Electronics and SK hynix target the 1c/1d DRAM generation; Micron targets the 1δ generation.
In plain terms = whether the 12-inch mask reaches pilot production on schedule in 2031 directly determines if all of these advanced-node roadmaps can be met — this is not one company's milestone but the industry's critical gate.
市场有风险，内容仅供研究参考，不构成投资建议。