Coherent Corp

Phase A — Understand the business

Lens 1 · Company Overview

Coherent Corp is a vertically integrated compound-semiconductor and photonics company — the merged entity of legacy II-VI (the acquirer, which adopted the "Coherent" name) and legacy Coherent Inc., combined in a ~$7B+ deal closed July 2022. It makes the optical engines, lasers, and exotic materials that move and shape light. Today the business is overwhelmingly an AI-datacenter optics story bolted onto a legacy industrial laser + materials base.

Plain-terms model: Coherent designs and manufactures (a) optical transceivers (800G / 1.6T pluggable modules) that connect GPUs, switches and servers inside AI datacenters; (b) the upstream components that go into those modules and into competitors' modules — indium phosphide (InP) lasers, EML/CW lasers, GaAs VCSELs, optical amplifiers, fiber-attach assemblies; (c) industrial lasers for materials processing, plus (d) compound-semiconductor materials including silicon carbide (SiC) substrates for power electronics. It is unusual in being both a module maker and a merchant component supplier to its own module rivals — it sells InP/EML lasers to the same transceiver market where Lumentum and the Chinese module houses compete.

Customers: hyperscalers and AI-system builders via the networking supply chain, with Nvidia now both a strategic supplier-relationship and a $2B equity holder (March 2026); Apple under a multiyear VCSEL/3D-sensing agreement; auto/power-electronics OEMs for SiC (the SiC subsidiary is backed by $1B from Denso + Mitsubishi Electric). Competitors: Lumentum (LITE), Fabrinet (FN, a contract manufacturer), Marvell (MRVL, a DSP/silicon-photonics component rival), and the Chinese module leaders Innolight (Zhongji) and Eoptolink. Contract structure has shifted decisively toward long-term agreements (LTAs) — management says orders now extend into calendar 2028 and customer LTAs run "to the end of the decade", a structural improvement over the historically book-and-ship optical-component model.

Reporting structure changed FY2026: the old three-segment model (Networking, Materials, Lasers) was realigned into two segments — Datacenter & Communications, and Industrial. As of Q3 FY26, Datacenter & Communications is ~75% of revenue, Industrial ~25%.

Lens 2 · Supply Chain

Map (named stakeholders, upstream → end demand):

• Upstream inputs / raw materials: InP and GaAs substrate crystal growth (Coherent does this in-house — a key vertical-integration point); III-V epitaxy; SiC boule growth. Coherent operates its own crystal/wafer fabs — in Aug 2025 it began production on what it calls "the world's first 6-inch indium phosphide production platform" in Sherman, Texas and Järfälla, Sweden. This in-house InP is the chokepoint that matters: InP CW/EML lasers are the scarce input for all high-speed transceivers and CPO.
• Component layer (Coherent as merchant supplier): InP lasers, EMLs, CW lasers, VCSELs (GaAs), optical amplifiers (SOAs), fiber-attach. Coherent sells these to other module makers — i.e., it is upstream of Fabrinet and partly of Lumentum.
• Module layer (Coherent as integrator): 800G-DR4 / 1.6T-DR8 pluggable transceivers; a VCSEL-based 1.6T transceiver planned for 2H calendar 2026.
• Switch / system buyers: Nvidia (Spectrum-X Photonics, Quantum-X — CPO switches shipping 2H2026 / early 2026), Arista, Cisco, plus white-box ODMs.
• End demand: hyperscaler AI-factory build-out (Microsoft, Meta, Google, Amazon, Oracle, xAI, etc., indirectly through the system vendors).

Chokepoints / single-source dependencies: (1) EML/InP laser availability is the industry-wide bottleneck — the search corpus explicitly notes Fabrinet's throughput is gated by "upstream EML laser availability regardless of its own capacity". Coherent owning InP capacity is its single biggest structural edge. (2) The Nvidia relationship is a concentration risk and a moat — Nvidia's $2B equity stake + multi-year laser-component purchase commitments + "future capacity access rights" + support for US fabs effectively makes Coherent a preferred CPO laser supplier, but ties a large share of the growth story to one buyer's roadmap. Names or it didn't happen: the chain is Coherent InP fab (Sherman TX / Järfälla SE) → Coherent lasers/EMLs → Coherent OR Fabrinet/Lumentum modules → Nvidia/Arista/Cisco switches → hyperscaler AI clusters.

Lens 3 · Competitive Advantages (moats)

Coherent's moat is vertical integration in compound semiconductors, not module assembly. Four durable advantages:

1. In-house III-V materials (the real moat). Owning InP and GaAs crystal growth + 6-inch InP scale-up means Coherent is not gated by the merchant laser shortage that constrains pure module assemblers. In a market where "EML laser availability" is the bottleneck, being your own laser supplier is the scarce capability. This is a process + capital-intensity moat that took decades and the II-VI materials heritage to build.
2. Breadth of portfolio. "One of the broadest portfolios in the industry, spanning lasers, optical components, networking equipment, and materials science," reportedly ~7,346 global patents. (Treat the exact patent count as web-sourced, unverified.) Breadth lets Coherent attach to whichever optical architecture wins — pluggable or CPO — because it supplies the photonic primitives either way.
3. CPO supplier lock-in via Nvidia. Nvidia's $2B investment + multi-year CW-laser supply agreement and capacity-access rights is a switching-cost + preferred-supplier moat specifically for the co-packaged-optics transition. CPO needs external high-power CW InP lasers — exactly Coherent's specialty.
4. Scale + customer LTAs. Orders booked into 2028 and LTAs "to the end of the decade" convert a historically cyclical component business into something with revenue visibility rivals lack.

Bargaining power: improving but asymmetric. Against customers — Coherent has gained pricing/allocation power because InP laser capacity is scarce and demand is backed out to 2028 (it can ration). Against Nvidia specifically — weaker; Nvidia is simultaneously a $2B shareholder, a key customer, and a buyer that also funded Lumentum ($2B) and is working with Marvell and Corning, so it is deliberately multi-sourcing. The moat is real but the most important counterparty holds structural leverage.

Lens 4 · Segments

No segments.csv exists (research layer empty) — all figures ``, point-in-time.

New two-segment structure (FY2026):

• Datacenter & Communications: Q3 FY25 $969M → Q3 FY26 $1,362M, +41% YoY; ~75% of total revenue. This is the growth engine — AI datacom is accelerating (Networking grew 56% YoY in an earlier FY25 quarter; datacom grew 79% YoY in Q3 FY25; +41% YoY in Q3 FY26 off a much larger base). Deceleration in the growth rate is purely the law of large numbers — absolute dollars are still ramping hard, and the mix shift toward 1.6T (higher ASP) supports margin.
• Industrial: ~25% of revenue, roughly $237M in a recent quarter for the old Materials line. This segment has been the laggard / cyclical drag — industrial/materials-processing demand was soft through 2024–25, with management calling only "early signs of an industrial uptick" by FY26. Portfolio pruning (aerospace & defense sale for $400M, Rofin-Baasel sale to Bystronic) is shrinking the non-core industrial footprint deliberately.

Geography: not reliably sourced here (no research-layer geo split). Coherent has meaningful China exposure on both the demand and competitive side; flag as an open item — n/a for a clean geo breakdown.

Trend read: the company is mix-shifting from a diversified industrial/materials conglomerate into an AI-optics pure-play, by both organic growth (datacom up, now 75% of revenue) and active divestiture of slower-growth, lower-margin units. That mix shift is the single most important driver of the margin expansion seen in Phase B.

Phase B — Measure performance

Lens 5 · Earnings Result

Latest print: Q3 FY2026 (quarter ended March 31, 2026), reported May 6, 2026:

• Revenue $1.81B, +21% YoY (+27% YoY pro forma adjusting for divestitures) — a record.
• GAAP gross margin 37.7% (+243 bps YoY); non-GAAP gross margin 39.6% (+105 bps YoY).
• GAAP EPS $0.97 (+$1.08 YoY — i.e., swung from a loss); non-GAAP EPS $1.41 (+$0.50 YoY, ~+55%).
• Datacenter & Communications +41% YoY drove the beat; Industrial improving off a soft base.
• Record bookings: "another step-function increase in the order book," backlog at a record, orders into calendar 2028, LTAs to end of decade.
• Q4 FY26 guidance: revenue $1.91B–$2.05B; non-GAAP GM 39%–41%; non-GAAP opex $360M–$380M; tax 18%–20%; non-GAAP EPS $1.52–$1.72.

Full-year context — FY2025 (ended June 30, 2025): revenue $5.81B, +23% YoY; GAAP GM 35.2% (+424 bps); non-GAAP GM 37.9% (+358 bps); GAAP EPS −$0.52 (a loss, but improved $1.32 YoY); non-GAAP EPS $3.53 (+$2.32 YoY, from $1.21); Adjusted EBITDA $1,350.4M, +35% YoY; ~$437M debt repaid in the year. Q2 FY2026 (Feb 4, 2026): revenue $1.69B +17% YoY; GAAP GM 36.9%, non-GAAP GM 39.0%; GAAP EPS $0.76 (+71%), non-GAAP EPS $1.29 (+35%).

Trajectory read: revenue +21–23%, gross margin climbing ~100–250 bps/quarter on mix + scale, and a swing from GAAP losses to GAAP profitability within FY2026 (Q2 GAAP +$0.76, Q3 +$0.97). The non-GAAP-to-GAAP gap (driven by amortization of the II-VI merger intangibles and SBC) is narrowing but still material — see Lens 10.

Notable vs own history: the stock has fallen on strong prints twice in the last year — down ~2% then rebounding on the FY25 Q4 beat (Aug 2025), and −23% in one session on Q2 FY26 when the otherwise-strong Q2 was overshadowed by Q3 revenue guidance the market read as light. The tape punishes guidance, not the print (see Lens 8).

Lens 6 · Earnings Calls (sentiment trend)

No transcripts/ in the research layer — sentiment read is `` from secondary coverage of the last ~4 calls (FY25 Q4 Aug 2025 → FY26 Q1/Q2/Q3).

Tone arc: steadily more confident, increasingly AI-singular. Under Jim Anderson (CEO since June 2024) the call narrative has converged on a "speed-to-market" + "AI datacenter" message, with management progressively raising the visibility horizon: from "strong demand" → "demand booked through 2027" → by Q3 FY26 "orders into calendar 2028" and "customer long-term agreements to the end of the decade". Recurring phrases: AI datacenter demand, indium phosphide capacity ramp, 1.6T, CPO, margin expansion, deleveraging, speed-to-market. CFO Sherri Luther's recurring theme is deleveraging — leverage 1.7x in Q2 FY26, down from 2.3x a year earlier.

What they stopped saying: the defensive "portfolio review / synergy capture / integration" language that dominated the immediate post-merger II-VI period has faded, replaced by offensive capacity-and-demand language. The aerospace/defense and Rofin divestitures are framed as focus, not retrenchment. Sentiment risk: the gap between bullish call tone and the −23% Q2 FY26 reaction shows the market is pricing perfection — management optimism is no longer a positive surprise, so guidance misses get punished hard.

Lens 7 · Comps

Peer set: pure-play / near-pure-play AI-optics. Multiples are ``, point-in-time mid-2026, and several are not cleanly sourced — flagged n/a rather than fabricated.

Read: COHR at ~49x forward / ~170x trailing GAAP is priced as a high-growth AI-infrastructure name, not a cyclical optics component maker. Lumentum's trailing ~159x P/E tells the same story sector-wide — the whole optical-component complex re-rated violently in 2026 (COHR +373% over the trailing year; 52-week range $76.88–$440.00). The honest comp conclusion: COHR is not obviously cheap or expensive versus its direct peers because the peers are equally extended; the relevant question (Lens 11/12) is whether the group's embedded growth is real. I will not invent EV/Sales or peer forward multiples I cannot source — the table above is deliberately sparse where the data isn't clean.

Lens 8 · Stock-Price Catalysts (>5% moves, last ~5 yrs)

All ``:

• Jan–Mar 2021 — bidding war. As legacy Coherent Inc., the stock was the target of a three-way war (Lumentum $100 cash + 1.1851 sh → II-VI $130 cash + 1.3055 sh → final II-VI $220 cash + 0.91 sh, ~$7B+). Large repricing events on each bid.
• Jul 2022 — merger close. II-VI becomes Coherent; balance sheet loads to ~$4.2B+ long-term debt.
• Jun 3, 2024 — Jim Anderson named CEO. Stock +~23% on the day. A management-quality re-rate.
• Aug 13, 2025 — FY25 Q4. Beat on EPS but −17% to −23% intraday at one point on the print/guidance read; later coverage shows it closed only ~−2% and rebounded after hours. Volatility on guidance interpretation.
• Feb 4, 2026 — Q2 FY26. Strong Q2 but −23.2% on light Q3 revenue guidance.
• Mar 2, 2026 — Nvidia invests $2B. Stock +~11.9% on the equity + CPO supply deal.
• May 6–7, 2026 — Q3 FY26. Strong print, but the optics trade cooled — COHR −10%, LITE −7%, AAOI −14% in a sector-wide pullback — then +16% on Jun 2 as the optics rally broadened back, and +7.48% on Jun 17 to ~$382.81.

Pattern the market actually reacts to: (1) forward guidance >> reported results — COHR repeatedly sells off on strong prints when guidance disappoints, the classic "priced-for-perfection" signature; (2) Nvidia / CPO headlines move it sharply (the $2B stake, CPO timing); (3) sector beta — it trades as a high-beta basket with LITE/AAOI on the "AI optics" theme, amplifying both directions. This is a momentum/expectations stock, not a value stock.

Phase C — Judge people & books

Lens 9 · Management

CEO Jim Anderson (since June 3, 2024). The signal is strong:

1. Track record — quantified and real. CEO of Lattice Semiconductor 2018–2024, where the stock rose ~10x with record operating profits and gross margins; prior SVP/GM of AMD's Computing & Graphics group; earlier Intel, Broadcom/Avago, LSI. He is a semiconductor-margins operator dropped into a materials/photonics company that had under-earned its asset base — exactly the right archetype for a margin-expansion + portfolio-focus turnaround.
2. Tenure & skin in the game — mixed. 2 years in. Directly owns 149,214 COHR shares ($60M at current price), bought 1,500 shares in the past year, sold none. As a % of the company that's tiny (0.013%), but he carries large legacy AMD ($246M) and Lattice ($90M) stakes, so his net worth is heavily semiconductor-equity-linked and aligned with the sector. Buying (even small) and not selling is a positive tell.
3. Capital allocation — early but coherent. Two clear moves: (a) deleverage — ~$437M repaid in FY25, leverage 2.3x→1.7x YoY; (b) focus the portfolio — sold aerospace & defense ($400M) and Rofin-Baasel, redeploying proceeds to debt + InP capacity. He's also secured the Nvidia $2B + Apple VCSEL + Denso/Mitsubishi $1B SiC capital — bringing in strategic capital to fund the InP/CPO ramp without over-levering. Coherent (the consensus picture) is reinvesting into the scarce bottleneck (InP) while shrinking the cyclical tail. ROE/ROIC are not cleanly sourced and were GAAP-negative in FY25 — n/a for a clean ROIC trend.
4. Red flags. The big one is compensation: Anderson was reported as the highest-paid CEO in America for 2024 at ~$101M (overwhelmingly stock awards tied to the hire). Defensible as a one-time inducement/turnaround grant, but it is a governance flag — a $101M package against ~0.013% direct ownership means the alignment is via granted equity, not bought equity. Watch dilution and the vesting structure.
5. Founder vs professional manager: a professional turnaround operator, not a founder. For a post-merger integration + margin story, that's the right fit; it implies execution and capital discipline over visionary risk-taking.

CFO Sherri Luther anchors the deleveraging narrative credibly. Net: above-average management quality, with the comp package the one thing a skeptic should keep flagging.

Lens 10 · Forensic Red Flags

Grounding caveat: no filings in the research layer, SEC EFTS down at fetch time, so this is a ``-grounded forensic read, not a 10-K line audit. Treat as a flagging exercise, not a clearance.

Income-statement / accounting risks to scrutinize:

1. The GAAP↔non-GAAP gap. FY25 GAAP EPS −$0.52 vs non-GAAP +$3.53 — a ~$4 swing. The wedge is dominated by amortization of merger intangibles (II-VI/Coherent purchase accounting) and stock-based comp (amplified by Anderson's $101M-skewed grant). Non-GAAP is doing a lot of work; a bull leaning on "$3.53 EPS" is leaning on adjusted numbers. The gap is narrowing as GAAP turned positive in FY26 (Q2 +$0.76, Q3 +$0.97), which is the right direction — but SBC dilution flatters non-GAAP and should be tracked against share count.
2. Goodwill / intangibles. A ~$7B merger leaves a large goodwill + intangible balance on a ~$14.9B total-asset base (total equity ~$5.64B as of 2025). Any datacom-demand air-pocket raises impairment risk. Not currently flagged, but it's the balance-sheet fault line.
3. Inventory / receivables vs revenue. With demand booked to 2028 and a fast InP capacity ramp, watch whether inventory builds ahead of revenue (capacity-ramp working capital) — a normal feature of this phase but a place where a demand stumble would show first. Not sourced cleanly here → open item.
4. Leverage, now improving. Net debt/equity 27.9%, leverage 1.7x (from 2.3x). Post-merger debt ($4.2B peak) is being worked down — a de-risking trend, not a red flag, but it constrained the equity story for two years and any rate/demand shock re-tightens it.

Regulatory findings (required sub-section).

• SEC Litigation Releases / AAERs: regulatory/regulatory-findings.md returned 0 SEC findings — BUT the EDGAR EFTS endpoint returned HTTP 500 at fetch time (2026-06-17), so this is not a confirmed clear — it is an unconfirmed/unavailable check. Re-run fetch-regulatory-findings.ts when EFTS is back before relying on it.
• Non-SEC enforcement: no material FTC/DOJ/FDA/CFPB action surfaced in the web pass. Note the optics supply chain carries US–China export-control exposure (compound semis, AI-datacenter components) — a policy/regulatory risk vector worth monitoring even absent a specific action.
• 10-K Item 3 (Legal Proceedings): not retrieved (no cached filing; ingest-filing.ts skipped per wave constraint). Open item — read the FY2025 10-K Item 3 on the next non-wave pass.
• Conclusion: No material regulatory or legal findings surfaced, but the SEC EFTS check failed (HTTP 500) and the 10-K Item 3 was not read — so this is "no findings surfaced, verification incomplete," not a clean "verified clear." Resolve on a non-wave refresh.

Phase D — Project & stress-test

Lens 11 · Forward Projection (no forecast.ts logged — unattended breadth loop)

Build from latest actuals + guidance. FY ends June 30. FY2026 is nearly complete (Q1–Q3 reported; Q4 guided).

FY2026 (base, near-actual):

• Revenue: Q1≈$1.5B + Q2 $1.69B + Q3 $1.81B + Q4 guide-mid $1.98B = **$6.98B**. Call it ~$7.0B, ~+20% YoY vs FY25 $5.81B.
• Non-GAAP EPS: H1 (Q1+Q2) ≈ $1.29 + $1.20 + Q3 $1.41 + Q4 guide-mid $1.62 = **$5.50 non-GAAP EPS FY26**. (Reconciles with ~49x forward P/E on ~$7.8 FY27 → see below: $382.81 / ~$7.8 ≈ 49x.)

FY2027 (the year the multiple is really pricing):

• Industry growth: AI-datacom transceiver TAM still expanding fast; 800G+1.6T forecast to exceed all other datacom transceiver types combined within ~5 years. Backlog booked into CY2028 de-risks the top line.
• Share / mix: 1.6T ramp + CPO laser content (Nvidia) raises ASP and mix; InP capacity (6-inch Sherman/Järfälla) is the supply unlock.
• Revenue base: Street consensus $7.05B–$9.45B for FY27. Base $8.4B (+20%), Bull $9.4B (+34%), Bear $7.1B (+1–2%, datacom digestion).
• Margins: non-GAAP GM guided to 39–41% and trending up ~100–250 bps/qtr; assume base non-GAAP GM ~41%, operating leverage on flat-ish opex.
• FY2027 non-GAAP EPS: Base ~$7.80; Bull ~$10.50; Bear ~$5.50–$6.00.

FY2028 (visibility year — orders already booked into CY2028):

• Base non-GAAP EPS ~$10.00; Bull ~$14 (CPO scale-up revenue from 2H2027 inflects); Bear ~$6 (CPO cannibalizes pluggable faster than Coherent's component content offsets, margin compresses).

Brier forecast: NOT logged (per --watchlist unattended rule; forecast.ts create skipped). Candidate if promoted later: "COHR FY27 (ends Jun-2027) non-GAAP EPS ≥ $7.50, p≈0.55."

Read: at $382.81 the stock is ~49x base FY27 EPS and ~38x base FY28 EPS. That requires the base (20%+ growth, 41%+ margins) to land and CPO to convert pluggable risk into component upside. It is priced for the bull-leaning base, with little margin for a datacom digestion year.

Lens 12 · Bull vs Bear

Bull case. Coherent is the vertically integrated arms dealer of the AI-optics supercycle. It owns the scarce input — InP/EML/CW laser capacity — that gates the entire transceiver and CPO market, just scaled it to a unique 6-inch InP platform, and is the partner Nvidia chose with a $2B equity stake + multi-year laser-supply + capacity-access deal. Demand is contractually booked into CY2028 with LTAs to the end of the decade — visibility no cyclical-components investor has had before. Under a proven semiconductor-margins CEO (Lattice 10x), gross margin is climbing ~100–250 bps/quarter, GAAP has turned positive, and leverage is falling 2.3x→1.7x. As 1.6T and then CPO scale (CPO scale-out revenue from 2H2026, scale-up from 2H2027), Coherent captures rising content per switch whether the industry ships pluggables or co-packaged optics — it supplies the photonic primitives either way. Earnings surprise lever: InP supply scarcity + ASP-rich 1.6T mix could push margins past the 41% guide. The bull number is ~$10.50 FY27 / ~$14 FY28 non-GAAP EPS.

Bear case (permanent-impairment risks).

1. CPO disintermediation, mis-timed. Co-packaged optics "eliminate pluggable transceivers". Coherent's module business (a large share of today's datacom revenue) is structurally threatened; the bull says the component (CW laser) content offsets it, but that's a lower-revenue, possibly lower-margin position, and the transition timing is Nvidia's to dictate. If CPO ramps faster than Coherent's component content compensates, datacom revenue and margin compress simultaneously.
2. Nvidia concentration + deliberate multi-sourcing. Nvidia funded Coherent and Lumentum ($2B each) and works with Marvell and Corning. Coherent is one of four bets; Nvidia is structurally engineering away single-supplier dependence. A roadmap change or allocation shift toward Lumentum impairs the thesis.
3. Expectations baked into the price. ~49x forward / ~170x trailing GAAP, +373% in a year, and the stock has fallen ~23% on a strong print (Q2 FY26) on light guidance. The market demands acceleration; a single in-line-but-not-blowout quarter de-rates a name this extended.

Pre-mortem (18 months out, thesis broke): Q4 CY2027 — hyperscaler AI-capex growth decelerates (a digestion year), CPO ramps slower-but-cannibalizes pluggables faster than expected, 1.6T pricing competes as Chinese module houses + Lumentum add InP-adjacent capacity, and Coherent prints two in-line quarters with no acceleration. The ~49x multiple compresses to ~25x on a flat EPS year → the stock halves even with flat fundamentals. The Nvidia stake caps the downside narrative but not the multiple.

Contrarian view (what the market refuses to see): The bull and bear both over-index on the module/transceiver narrative. The durable story is the III-V materials monopoly position (in-house InP at 6-inch scale) — that's the asset that survives whether pluggable, LPO, or CPO wins, and it's the part of Coherent least visible in a transceiver-unit forecast. The market is trading COHR as an optical-module beta basket with LITE/AAOI; the differentiated truth is that it's a compound-semiconductor materials company that happens to also make modules — and the materials moat is what justifies a premium the module business alone would not.

Lens 13 · Devil's Advocate (short-seller)

I'm short. Here's the dismantling:

• Revenue concentration is worse than it looks. ~75% of revenue is now Datacenter & Communications, and that is concentrated into a handful of hyperscaler end-buyers through a handful of system vendors, with Nvidia as the gravitational center — a single counterparty that is also a shareholder and is explicitly funding my competitors. Concentration into one buyer's roadmap is not a moat; it's a leash.
• The moat may be weaker than bulls think. "Vertical integration in InP" is real today, but every dollar of AI-optics profit is a magnet for capacity: Lumentum (also $2B Nvidia-funded), the Chinese module leaders (Innolight + Eoptolink already ~60% of 800G), and Nvidia's own silicon-photonics + Corning glass effort. InP scarcity is a 2026 condition, not a permanent law of physics — capacity catches up, and when it does the pricing power evaporates and COHR re-rates to a cyclical multiple.
• Most dangerous competitor bulls underestimate: Nvidia itself. Nvidia is vertically integrating optics (Spectrum-X/Quantum-X CPO, silicon photonics, four funded suppliers). Coherent's best case is "preferred laser supplier to Nvidia"; the bear case is "commoditized laser supplier to a vertically integrating customer that funded three alternatives."
• Capital-allocation / governance flag: a $101M CEO pay package (highest in America, 2024) against ~0.013% bought ownership — alignment by grant, not conviction. Plus a non-GAAP EPS ($3.53) that is ~$4 above GAAP (−$0.52) on merger-intangible + SBC adjustments. The headline earnings power is an adjusted construct.
• What must hold for $382.81: 20%+ revenue growth through FY28, margins to 41%+, CPO converting from threat to content-tailwind, and no hyperscaler-capex digestion. That's four things, all going right, with the most important counterparty (Nvidia) holding the leash.
• Growth disappoints 20–30%: if FY27 lands at the bear $7.1B (consensus low) instead of base $8.4B, non-GAAP EPS is ~$5.50–$6.00, and a de-rate to ~25x → ~$140–150 stock, ~−60%. A momentum name with a 52-week low of $76.88 has demonstrated it can trade there.
• Single permanent-impairment scenario: CPO standardizes on an architecture where Nvidia/partners in-source the laser or dual-source it to Lumentum at commodity prices, and AI-capex normalizes — Coherent's datacom revenue stops growing and the materials premium never gets recognized. Plausibility: medium-low on a 1–2yr view (backlog to 2028 protects near-term), rising on a 3–5yr view.

Lens 14 · Management Questions (ordered by information value)

1. As CPO ramps (scale-out 2H2026, scale-up 2H2027), what is your content-per-switch in dollars under CPO vs today's pluggable module, and does the gross-margin dollar per AI cluster rise or fall as the industry transitions?
2. Quantify Nvidia: what % of FY26 and backlog-to-2028 revenue is tied to Nvidia (direct + through its switch ecosystem), and how do you protect ASP given Nvidia also funds Lumentum and works with Marvell/Corning?
3. How much of the CY2028 backlog is take-or-pay / non-cancelable vs forecast/LTA-framework that can be revised down if hyperscaler capex digests?
4. What is the InP capacity ramp curve (6-inch Sherman + Järfälla) in wafers/quarter through FY2028, and at what utilization does it become margin-accretive vs the current ramp drag?
5. Walk through the path to closing the GAAP↔non-GAAP gap: when do merger-intangible amortization and SBC normalize enough that GAAP EPS approximates non-GAAP?
6. What is normalized free cash flow conversion at scale, and what is the capex intensity required to hold the InP supply lead through the CPO transition?
7. What is your deleveraging target leverage ratio and timeline, and what's the priority for capital once you're there — buybacks, M&A, or further InP/SiC capacity?
8. Where do you see 1.6T (and 3.2T) pricing going as Chinese module houses and Lumentum add capacity — what's the ASP erosion curve, and how does in-house InP defend it?
9. What is the long-term plan for the SiC business (Denso/Mitsubishi-backed) — is it core, or a candidate for the same focus-driven divestiture logic as aerospace/defense and Rofin?
10. What share of datacom is merchant component sales to competitors (selling InP/EMLs to module rivals) vs your own modules — and how do you manage that channel conflict as CPO shifts the value pool toward components?
11. What is the realistic LPO vs CPO vs pluggable revenue mix you're underwriting for FY27/FY28, and what happens to your model if CPO adoption is 12 months faster or slower than planned?
12. How exposed is the supply chain and customer base to US–China export controls on compound semiconductors and AI-datacenter components?
13. What is the Industrial segment's normalized growth and margin once the cyclical recovery completes — is it a structural drag worth shrinking further, or a diversifier worth keeping?
14. On capital allocation: how should investors read a ~$101M CEO grant against ~0.013% open-market ownership — what is the vesting/performance structure, and would you commit to meaningful open-market purchases?
15. What single internal metric do you watch that would tell you the AI-optics demand cycle is rolling over before the order book does?