Optical Supply Chain Scan — October 16, 2026

Evidence as of 2026-10-16

Settled October 16 tape: the 28-name optical list was 0 of 28 up on the week and on Friday, and among this campaign's five names only Lumentum holds above its 200-day average. Coherent, Ciena and Credo are long trends under test; Fabrinet is negative over one year with negative alpha. Price action is not supplier evidence and fires neither failure leg; no issuer statement on file explains the move.

KawzInvests — Celestica's OCP data: transceiver qualification windows shrink to 18 months while GPUs refresh every 12

Convequity — Lumentum's laser demand rides on CPO design choices that are still open

Evidence as of 2026-10-09

Convequity argues Lumentum's external-laser and optical-switch demand depends on co-packaged-optics and switching designs that are still open, and that on-package light sources, micro-LED short links and LiDAR-derived silicon photonics could take share. That names the substitution channel in the first failure test, but it is an analyst's opinion with no order, revenue or margin data, so it is context rather than a challenge. The test remains two supplier reports showing order weakness with margin deterioration.

Serenity — Lumentum's CEO says the company can supply only 30% of 2027 demand as co-packaged optics arrive

Evidence as of 2026-10-02

A third-party X transcription (Serenity) of Lumentum's CEO at the Global Photonics Economic Forum: 2027 supply covers about 30% of demand, with balance by 2029-2030. This is management guidance about allocation, not a Lumentum release, filing or transcript, and not a delivered result. It is consistent with demand absorbing capacity but establishes no orders, revenue, margin or price terms. The CEO's later Bloomberg remark (sold out through early 2029, relayed on X October 8-9) is the same class and has no primary source on file. Test both against the November 5 report.

Optics Is Pricing the Laser Shortage, Not the Module Cycle

KawzInvests — Credo's growth beyond copper: optics above $600M in FY27 plus a 1.6T AEC ramp

Serenity — Morgan Stanley hears 3.2T Chinese transceiver curbs are coming, with a 65% US-content escape hatch

Vik's Newsletter — ECOC 2026 product announcements around optical scale-up

Serenity — Foxconn's interconnect chairman names CW lasers and fiber the photonics bottleneck

Terry — Citi Global TMT readout: NAND TAM $50B to $500B by 2027, AMD CPU TAM $60B to $220B, optical testing capacity about 100x short

Campaign 1: INTACT, provisional

Evidence as of 2026-09-04

Earlier same-day thesis review retained as context. Its prior classification is historical; the current scope review supplies the live disposition and preserves its unresolved evidence gaps.

Coherent: optical growth and capacity expansion

Evidence as of 2026-08-12

Supplier operating progress supports optical demand converting into economics. Allocation duration, capacity response and margin persistence still require follow-through; outlook is not delivered growth.

CoreWeave: delivered growth and financing demands

Evidence as of 2026-08-11

AI-cloud expansion is buyer-side context. Supplier disclosures, rather than a buyer’s backlog, determine optical allocation and margin evidence.

Lumentum: optical demand and product ramps

Evidence as of 2026-08-11

Supplier operating progress supports optical demand converting into economics. Allocation duration, capacity response and margin persistence still require follow-through; outlook is not delivered growth.

MPS: power conversion reaches customers

Evidence as of 2026-07-30

Power products inside optical modules corroborate an architectural demand channel, but do not establish optical-supplier allocation or margins.

Serenity - Broadcom's CEO: laser demand "far surpassing supply"

Photon Capital — COHR’s AI mix and InP capacity ramp

Viks Newsletter — where copper and optics still contradict each other

Gaetano (crux_capital_) — the market sold off the wrong optical name: his bull case for why AXTI's China selloff is negotiating leverage, not a real ban

Serenity (aleabitoreddit) — AAOI's earnings call says China is years behind on CPO lasers, and AAOI itself is capacity-out of first-gen CPO

ChurchillWw — ZEISS is expanding all three of its main German optics plants at once

AXT — Tongmei withdraws Shanghai STAR IPO bid, pivots to Hong Kong listing for InP business

crux_capital_ — the photonics buildout isn't over, even at the height of fear

Fifteen of twenty broke together

Disclosed money: Congress's most active traders spent May and June buying the AI supply chain

Nutty — Investor's Handbook III: Substrate, where glass packaging starts getting paid

Viks Newsletter — Lasers for CPO/NPO, Part 1: how the InP DFB laser works and where it hits a wall

The Unwind Is Washing Out Our Picks-and-Shovels Lanes While Their Fundamentals Inflect Up

aibottlenecks snapshot — 2026-07-02 (deltas vs 2026-06-04 REST / 2026-06-03 DOM baseline)

@crux_capital_ — the two real bear cases for the photonics supercycle (laser oversupply vs VCSEL-over-SiPh)

Investigation — Post-beat-decline pattern across optical-supercycle Wave 1

Investigation — Aschenbrenner essay vs 13F scorecard — what the Situational Awareness predictions have actually done, and the new megacap put overlay

Investigation — Aschenbrenner research dossier — essay-mined lanes, the pair-trade put book, 13F track-record calibration, and cross-manager convergence

Investigation — SpaceX S-1 picks-and-shovels / supply-chain investable map

aibottlenecks.app monthly snapshot — 113-name chokepoint universe baseline — dashboard-snapshot

Astera Labs ALAB retimer thesis — most boring chip highest gross margin in the rack — tweet

AI-Lab S-1 → Thesis Crosswalk Playbook

Investigation — xAI (SpaceX AI segment) S-1/A crosswalk read — first application of the AI-lab S-1 thesis crosswalk playbook

CIEN — Deep Dive

$AEHR — Deep Dive (Aehr Test Systems, wafer-level burn-in)

Evidence as of 2026-05-18