Structural shift
Registered claim
Indium phosphide (InP) substrate capacity emerges as an explicit named constraint on AI compute interconnect deployment by 2026-Q3, on par with CoWoS-as-named-constraint in the prior 2024-2025 cycle. Pattern is established by 5+ independent attestations (VECO + AXTI + GFS SCALE + GLW Springboard + IntelliEPI primary-supplier voice) and at least one hyperscaler will explicitly cite InP supply as a deployment constraint in Q2 or Q3 2026 commentary.
Registered 2026-05-10.
Resolution spec
Spec registered 2026-06-12.
- [A] CORE :: At least one hyperscaler explicitly cites InP supply as an AI-compute deployment constraint in Q2 or Q3 2026 public commentary. This is the prediction's specified resolution surface - hyperscaler citation; supplier-side attestation differs in kind and does not score this leg (read-through discipline, catalyst-file-generation spec section 5 invariant 2).kind: dated binary event :: event date 2026-09-30 :: evidence: primary source
- [B] confirm :: InP capacity constraint attested by 5+ independent supply-chain voices (distinct issuing entities).kind: attestation count :: threshold 5 :: independence: distinct issuing entityattested: Veeco (2026-05-06)attested: AXT (2026-05-06)attested: GlobalFoundries (2026-05-06)attested: Corning (2026-05-06)attested: IntelliEPI (2026-05-10)attested: Coherent Corp (2026-06-18)attested: AXT Inc (2026-06-18)attested: Lumentum (2026-08-12)
Verdict rule: HIT if leg A and leg B. PARTIAL if not leg A and leg B. MISS if not leg B.
As frozen: HIT = A && B; PARTIAL = !A && B; MISS = !B
Tier history
- 2026-05-15: MEDIUM (initial). Trigger: Hold tier=medium within photonics frame. Direct supports accumulating - IntelliEPI explicit InP supply-shortage warning (0.80 ind x3), Veeco $250M+ InP laser equipment orders for 800G/1.6T transceivers (0.95 primary), AXTI Q1 InP revenue acceleration (0.60), AAOI first 1.6T volume order (0.85). But the prediction's framing - "explicit named constraint on par with CoWoS-as-named-constraint" - requires mainstream sell-side / industry-narrative recognition that has not yet materialized. Evidence-accumulation phase; not yet decisive for tier=high upgrade. Watch for sell-side reports explicitly framing InP supply on par with CoWoS over next 2-3 months.
- 2026-05-16: MEDIUM > HIGH. Trigger: Tier moves from medium to high within photonics frame. Today's InP-shortage-deepening datapoint is a direct SME-confirmation of the InP-capacity-as-named-constraint Q3 2026 prediction; AVGO-hidden-InP-laser-division datapoint strengthens the structural-beneficiary set (AVGO as #2 InP player after LITE). Insane Analyst rates InP as the #1 stack-layer bottleneck - single-source but with quantitative-direction (worsened over 6 months) and structural-specificity (substrate→wafer→epi→laser-printing chain). The 5-layer InP convergence pattern is reinforced.
- 2026-07-31: HIGH > MEDIUM. Trigger: The confirm leg is already banked - seven distinct supply-chain entities attest the InP constraint against a threshold of five - so the floor here is PARTIAL and the prediction cannot MISS. What changed is the core leg's prospects. Leg A requires a HYPERSCALER to cite InP supply as an AI-compute deployment constraint, and today AXT disclosed it is doubling InP capacity during 2026 ahead of schedule toward roughly 60M USD quarterly exiting 2026 and approximately 130M USD exiting 2027, funded by a 632M USD raise lifting cash to 749M USD. A constraint being resolved on a stated schedule by its own oligopolist is a constraint less likely ever to surface in hyperscaler commentary. cap-inp-epiwafer-substrate-oligopoly wobbled today and was narrowed to time-bound its availability limb for exactly this reason.
Evaluated daily since registration: 51 evaluations. The 3 that moved the tier are listed above.