Type: thesis-shift (subtype: hardens; sub-event continues 2026-05-06 NVDA $4B entry)

event_id: 2026-05-06-cohr-equity-instrument-axti-merchant-confirmation

Sources:

• Coherent 8-K filing (2026-03-02) — Item 3.02 (Unregistered Sales of Equity Securities), Item 7.01 (Regulation FD), Item 9.01. Discloses 7,788,161 shares of common stock at $256.80 = $2.0B, Section 4(a)(2) private placement to NVIDIA. NOT preferred.
• NVIDIA Newsroom (2026-03-02) — parallel announcement language: "future access and capacity rights for advanced laser and optical networking products."
• Coherent CEO interview (24/7 Wall St, 2026-03-11) — characterized commitment as "multibillion-dollar... extending through end of decade."
• Coherent Q3 FY2026 earnings transcript (Motley Fool, 2026-05-06) — date discrepancy with original task framing (5/13). Worth verifying whether the call has actually happened or this is a pre-publication artifact. Either way the structural facts come from the 8-K, not the call.
• Telecompaper (2026-03-26) — confirms Lumentum acquired Greensboro NC fab from Qorvo for $18M.
• AXT Q1 2026 earnings transcript (Motley Fool, 2026-04-30) — InP revenue $13.6M (>50% of total revenue); $100M record InP backlog (up from $60M Q4'25 / $49M Q3'25); "tier-one laser manufacturers" + "end hyperscalers" language.
• Lumentum-IQE multi-year strategic supply agreement extension (IQE press release, 2025) — confirms LITE merchant-epiwafer model.
• Mordor Intelligence InP wafer market structure — Sumitomo 30% incumbent, top-5 = 70%.
• optics.org (2026) — Lumentum's $43M quarterly "InP wafer fabrication facility" investment described as device-layer (epi + lithography), not LEC/VGF crystal pulling.

Shift: Two structural questions from the 2026-05-06 NVDA $4B entry now resolved.

(1) COHR vs LITE deal symmetry: SYMMETRY HOLDS on 4 of 5 elements; 1 material divergence. Both deals share the same playbook (multibillion purchase commitment + capacity-access rights + R&D collaboration + equity leg + anchor framing) but use different equity instruments. LITE got Series A Convertible Preferred at $695.31 with conversion economics and (presumably) liquidation preference. COHR got common stock at $256.80 — no conversion economics, no liquidation preference, no dividend stack. Common is structurally weaker downside protection. Read: NVDA negotiated less protection from COHR likely because (a) COHR's CW-laser/CPO-ingredient stack is broader/less single-threaded than LITE's transceiver-concentrated position, and (b) COHR's larger market cap and float was sufficient to absorb $2B common print without preferred mechanics. HBM-playbook mapping holds on both sides; equity-instrument asymmetry is a real signal worth tracking — LITE got the protected leg, COHR got the print.

(2) Greensboro captive-vs-merchant InP substrate question: MERCHANT-substrate, captive-device. Three convergent evidence lines: (a) $18M acquisition price vs AXT's own $220-250M greenfield estimate for a substrate-growing facility — it's not a crystal-pulling fab; (b) Lumentum's consistent passive "leveraging 6-inch InP wafers" language across announcements is consumption-side, not crystal-growth-side; (c) Lumentum already has multi-year merchant epiwafer agreement with IQE, confirming the outsource-the-substrate model. Coherent built its own 6-inch InP capability in Sherman TX + Järfälla SE separately — suggesting independent substrate capability would be a multi-year separately-disclosed capital project. Lumentum has disclosed nothing of the sort. AXT promoted from speculation tier (↑) to direct-beneficiary anchor name (↑↑↑). $100M record InP backlog with "tier-one laser manufacturers" landing exactly as Lumentum ramps Greensboro is not coincidence; substrate orders for 2028 production ramp NOW (lead time alignment).

Effect on thesis: HARDENS the parallel-deal symmetry framing AND adds AXTI as a direct-beneficiary anchor. Caveats remain: Sumitomo Electric is 30%-share incumbent with 6-inch capacity sooner than AXTI — AXTI's 6-inch transition is "longer-term" (~12+ months) while customers want 6-inch now, so Sumitomo likely captures the immediate Greensboro ramp with AXTI taking second-source / China-fed supply. AXTI also carries China export-license overhang (AXT Tongmei subsidiary needs MOFCOM permits, ~3-month cycle).

Per-ticker:

• COHR ↑↑ (parallel-deal symmetry confirmed; equity-instrument asymmetry is informational, not bearish — common print is the structurally weaker form but doesn't undermine the supply-lock thesis)
• AXTI ↑↑↑ PROMOTED from speculation to direct-beneficiary anchor. Not a watchlist change (AXTI was already in key_tickers); narrative status change. Re-rate from "potential promote-from-watchlist" to "direct-beneficiary, second-source structural play behind Sumitomo."
• LITE = (no change — already ↑↑↑; reinforced by COHR-asymmetry signaling LITE got the protected equity leg)
• SOI.PA = (no change — separate SOI substrate question; not the InP question)

Open questions:

• Is the Q3 COHR transcript dated 2026-05-06 a real call or a pre-publication artifact? Original earnings calendar said 2026-05-13. If real, Q3 color includes scale-out CPO ramps 2H 2026 + scale-up 2H 2027 + Sherman TX named as NVDA-partnership capacity node, but no specific NVDA dollar commitment beyond "multibillion" disclosed yet.
• 13F check next quarter — does NVDA's COHR common stock show up vs LITE preferred (which may not under SEC reporting treatment of convertible preferred)?
• 10-Q footnote check for LITE/COHR purchase-commitment dollar floor (commitments-and-contingencies disclosure should eventually quantify "multibillion").
• Lockup duration on the 7,788,161 COHR common shares — material for float dynamics.
• Coherent's Sherman TX 6-inch line — does Coherent grow its OWN 6-inch substrate in Sherman, or also buy from AXT/Sumitomo? If captive, hurts AXTI's TAM among the top-2 device makers.
• Sumitomo Electric capture share of the Lumentum/Coherent ramp — Sumitomo is 30%-incumbent and has 6-inch sooner; how much of the Greensboro ramp goes to Sumitomo vs AXTI second-source?

Methodology note: Picks-and-shovels paradigm bonus on AXTI — by tracing the substrate layer below the device fab, we're now positioned at the upstream chokepoint of the optical supercycle. This is the same shape as the HWM-vs-PCC framing on the IGT-blade thesis (alloy-layer below the engine OEM): once you find the merchant-supply layer below a captive device fab, you have a pure-play that benefits from total industry growth without picking the right device-maker.