ASML/Zeiss's EUV mirror advantage is attributed to embedded tacit know-how rather than patent protection, which Nikon and Canon have not overcome

Extreme-ultraviolet (EUV) lithography optics are described as reproducing a centuries-old pattern in precision-instrument economics: the competitive moat is embodied institutional skill, not intellectual property. On this account Nikon and Canon — both established precision-optics and DUV-lithography firms — "have been unable to enter EUV after 15+ years of industry awareness that EUV was the path forward," despite no patent wall stopping them. The barrier is characterized as "technical rather than commercial": the ASML/Zeiss mirrors are reportedly polished below 0.1nm RMS roughness through decades of joint institutional development that a competitor cannot simply copy.

[unverified-mechanism — needs primary ASML/Zeiss documentation or Tier 1-2 industry analysis] and [unverified-quant — the sub-0.1nm roughness figure and the "15+ years" timeline need a primary]. The entire claim, both its causal mechanism and its numbers, rests on a single Tier-4 industry-explainer (semiconductorx.com), which is below the sourcing floor for technical-mechanism and quantitative claims. It is recorded and flagged rather than dropped, and the note is held at seedling; the verification is queued at question-verify-euv-mirror-tacit-knowledge-moat.

2026-08-01 update: that question has since been answered. The roughness figure is corroborated at Tier 1 by Zeiss's own 2011 engineering data (claim-zeiss-2011-slide-deck-confirms-0-1nm-rms-euv-mirror-surface-figure). The general tacit-knowledge-not-patents thesis is independently confirmed at Tier 2 (claim-burja-and-ip-business-academy-source-asml-zeiss-euv-moat-to-tacit-know-how). But the "15+ years" framing directly above — Nikon and Canon persistently trying and technically failing — does not survive contact with the better-sourced history: the documented mechanism is exclusion from the EUV-LLC consortium, followed by Nikon's strategic exit after reaching a 2008 prototype (claim-nikon-canon-euv-exclusion-was-consortium-politics-not-failed-imitation). Treat the paragraph above as the claim this note originally made, not as settled history.

If it holds, the claim is the modern leg of a four-century structural parallel: paper-level secrecy fails (claim-lippershey-1608-telescope-patent-denied-as-too-easy-to-copy) while hands-and-institution secrecy endures (claim-spinoza-grinding-dish-outlived-him-in-huygens-hands) — the same tacit/embodied moat as Spinoza's grinding dish, now at fab scale. It also sits adjacent to the vault's export-control thread: know-how of this kind resists both patents and sanctions, unlike the transferable machine tools of claim-toshiba-machine-sold-ussr-submarine-quieting-tools, and unlike a diffusible published stock it does not self-liquidate the way the expertise in claim-soviet-expertise-was-finite-stock-depleted-as-it-diffused did.

The same tacit/non-codifiable property that makes this a moat against imitation makes the inverse problem in AI governance: when a load-bearing verdict cannot be codified it cannot be automated, so CANONIC falls back to a codifiable accountability ledger rather than a truth-gate — see claim-canonic-deliverable-is-an-append-only-evidence-ledger and the shared root (Polanyi's paradox / codification) in 2026-07-11-hop-polanyi-codification-bridge.

It is also a boundary case for Sahal's identity: that identity's own precondition — cumulative production growing exponentially in calendar time — fails badly for EUV mirrors (ASML's Zeiss partnership began in 1986, 24 years before the first EUV shipment in 2010, and cumulative EUV shipments took another decade to reach 100 units), so Wright's Law and Moore's Law should not collapse into one curve here. That is consistent with reading the "15+ years" persistence of the moat as calendar-time-accumulated institutional experience rather than a production-volume learning curve — see 2026-07-16-hop-euv-sahal-boundary.