Theses›Semiconductor manufacturing›Lithography and patterning

Lithography and patterning

ASML/EUV lithography is the #1 durable constraint on AI-compute scaling by ~2028-2030 (one supplier, more concentrated than memory).

  • Early
  • $55B by 2035

The call, a draftEUV is winning and stays the hardest constraint on AI chips through about 2030, with ASML the only supplier. No alternative tool reaches 2nm logic before about 2032. The tool layer is closed; the materials layer is early.

D1D2D3R1R2R3C1C2foragainst D1D2D3R1R2R3C1C2foragainst
Fig. 1The balance. Drivers and catalysts against restraints, each block sized by its weight; catalysts hang dashed because they have not happened yet, sized by their odds. It tips toward the drivers.
  1. D1ASML is the sole EUV supplier, and AI chip demand at 2nm and below runs through its High-NA tools from 2026 to 2030.
  2. D2Photon shot noise at High-NA is a physics limit for wet chemically amplified resists, which opens a slot for dry and metal-oxide chemistry; JSR qualified a metal-oxide resist on imec's High-NA pilot line in 2023 and Lam showed 28nm-pitch dry resist in Jan 2025.
  3. D3Canon's nanoimprint is limited to 14nm lines and no foundry has committed it to high-volume logic, so the alternatives stay specialty.
  4. R1The scanner is out of reach for any new company, and the resist market is small and held over 90% by three Japanese suppliers.
  5. R2A custom photoresist company forecast its own revenue topping out around $20M, a sign the resist slice may not support large independent businesses.
  6. R3Absolics' glass-core substrate slipped from end-2025 mass production to 2026 to 2027 and is still pre-qualification with AMD and AWS.
  7. C1A foundry qualifies a dry or metal-oxide EUV resist into a High-NA production node.31 Dec 2028
  8. C2A Western glass-core substrate line reaches AI-accelerator qualification with a named customer.31 Dec 2027
Rests onthe beliefs underneath
Supported by · confidence
What would prove itdated tests, open and resolved
202720282029203020312032today12345678predictionswatch 202720282029203020312032today12345678predictionswatch
Fig. 2Each ring is a dated test: dashed while open, struck through when overdue, ticked or crossed once resolved.
  1. 1At least one Western glass-core substrate line reaches AI-accelerator HVM qualification (named customer/product) by end-2027 — glass is the credible Western counter-bet.Dec 2027 · open
  2. 2A dry / metal-oxide EUV resist is qualified into a High-NA logic node by end-2028 — the MATERIALS layer crystallises the seed value, not a tool.Dec 2028 · open
  3. 3Lace (atom-beam) does NOT demonstrate >1 wafer/hour at <10nm by its ~2029 pilot-tool milestone — throughput, not resolution, is the wall.Dec 2029 · open
  4. 4Meta-call: the fundable European seed entry in the litho/patterning stack is enablement materials and data, NOT a litho/patterning tool or a substrate fab.Dec 2029 · open
  5. 5Lace disclosing wafers per hour below 10nm at its pilot tool.Dec 2029 · open
  6. 6ABF substrate stays >90% single-sourced from Ajinomoto with zero US/EU HVM production through end-2030, despite CHIPS/EU-Chips money — the gap is geographic, not just technological.Dec 2030 · open
  7. 7No alternative-patterning platform is named in any ≤2nm logic production roadmap by end-2031 — beyond-EUV stays specialty/NAND.Dec 2031 · open
  8. 8Any foundry naming a non-EUV patterning platform in a 2nm logic flow.Dec 2031 · open
Expressionslisted, private, technologies
Private · 11 private companies, 7 of them in the US.

Irresistible Materials · IMS Nanofabrication · Multibeam Corporation · Absolics · Averatek · Ajinomoto Fine-Techno · Eulitha · Inpria · Substrate · xLight · Obsidia Semiconductors

9 more private companies tracked, not shown

Technologies
What changedthe record, newest first
  • 23 Jun 2026The post-EUV glass-substrate timelines are slipping (Absolics now targets 2026/27, not 2025), while the dry and metal-oxide resist path keeps hitting dated milestones; the durable bet still sits in the materials, not the tool.
  • 17 Jun 2026Tier-1 chipmakers are actively pulling for 3D and non-planar patterning (curved optics, thick resist, advanced packaging), separate from the leading-edge EUV story; but a real demand signal still does not make a new patterning tool seed-investable, and the durable value stays in the resist chemistry, not the stepper.
  • 3 Jun 2026A wave of capital is flowing into post-EUV lithography tools, but the investable seed surface is still the materials and data underneath, not the tool.
Reading
Show the reading