INSTITUTIONAL EQUITY RESEARCH | AI INFRASTRUCTURE SERIES
Coherent Corp. (NYSE: COHR)
AI Photonics Infrastructure · Vertical InP Integration · 1.6T Transceiver Ramp · NVIDIA Strategic Partnership
June 2026 · CII Research · Manu · AI Infrastructure Series — P1
A. EXECUTIVE SUMMARY
|
Category |
Rating |
Commentary |
|
Price (approx.) |
~$350 |
S&P 500 entry ~$249 (Mar 2026); +40% since inclusion |
|
Market Cap |
~$67B |
Post-NVIDIA equity investment; total equity >$11B |
Core Investment Thesis
Coherent Corp. is the clearest pure-play institutional bet on the photonics layer of AI infrastructure. The company is not simply riding the AI wave — it is the physical substrate over which that wave travels. With the world's only 6-inch indium phosphide production lines, a vertically integrated manufacturing stack that competitors cannot replicate on any realistic timeline, and now a $2 billion NVIDIA strategic investment anchoring a multi-year supply relationship, Coherent has transitioned from a debt-heavy post-merger recovery story into something closer to a structural compounder in the making.
The NVIDIA deal is not just a supply agreement. It is a customer funding your capacity expansion in exchange for supply priority — structurally similar to how TSMC's hyperscaler customers commit to node capacity. This de-risks capex, validates the moat, and locks in revenue visibility.
B. BUSINESS MODEL & ECONOMIC MOAT
What Coherent Actually Does
Coherent (formerly II-VI Incorporated, which acquired Coherent Inc. in a $7B deal in 2022) is a vertically integrated photonics company. It designs, manufactures, and sells optical transceivers and components for datacenters (800G, 1.6T, and next-generation co-packaged optics), indium phosphide and silicon photonics-based lasers and chips, industrial lasers for precision manufacturing and semiconductor capital equipment, and Silicon Carbide substrates through a JV with Denso and MELCO.
Revenue Composition
|
Segment |
Rev Share (Q3 FY26) |
YoY Growth |
Outlook |
|
Datacenter & Communications |
75% |
+41% pro forma |
Strong — 1.6T ramp + CPO pipeline |
|
Industrial |
25% |
+2% |
Stable — steady cash flow base |
Economic Moat Assessment
|
Moat Source |
Strength |
Trend |
Confidence |
Reasoning |
|
Proprietary Technology (InP) |
Strong |
Strengthening |
Very High |
|
|
Vertical Integration |
Strong |
Strengthening |
High |
|
|
Scale Advantages (6-inch InP) |
Moderate |
Strengthening |
High |
|
|
Switching Costs (Hyperscaler qual.) |
Moderate |
Stable |
Medium |
|
|
Customer Lock-in (NVIDIA deal) |
Moderate |
Strengthening |
High |
|
|
IP Portfolio (7,346 patents) |
Moderate |
Stable |
Medium |
|
|
Network Effects |
Weak |
Stable |
Low |
|
|
Regulatory / Geopolitical Barrier |
Moderate |
Strengthening |
Medium |
The moat here is a process and materials moat — the kind Munger recognizes as deeply embedded in proprietary manufacturing know-how. Coherent's 20-year InP platform cannot be replicated in any realistic 3–5 year window. The 6-inch production line is the commercial expression: world-first, yields already exceeding 3-inch predecessors from day one.
C. COMPETITIVE MOAT DURABILITY — THE 10-YEAR TEST
Can Coherent earn excess returns on capital 10 years from now? Conditionally yes — if it successfully navigates the technology transition from 800G → 1.6T → 3.2T → CPO.
Why the Moat Strengthens (3–5 Years)
• Three simultaneous laser technology routes for 1.6T: silicon photonics (SiPh), InP EML, and GaAs VCSEL — competitors are typically committed to one route only
• 4x InP capacity expansion planned by end-2027 — scale further widens the unit economics gap
• US CHIPS Act tailwinds — domestic InP manufacturing is geopolitically strategic; government incentives reinforce CapEx economics
• NVIDIA CPO partnership positions Coherent for the next architectural generation, not against it
Where the Moat Faces Stress
• Chinese vendors (Innolight and others) are competing hard on module assembly — delivery capability and yield management are contested terrain
• Co-Packaged Optics (CPO) could disintermediate today's pluggable module business if Coherent misses design wins
• Hyperscalers have both incentive and leverage to keep module pricing in check — buyer power is concentrated
The 10-year verdict: The moat is conditionally durable. If Coherent transitions successfully from pluggable to CPO while maintaining InP laser supply dominance, it remains a critical chokepoint in AI infrastructure. If it misses CPO, the structural position weakens from 2028 onward.
D. ROIC & VALUE CREATION FRAMEWORK
Historical ROIC (Estimated)
|
Period |
Estimated ROIC |
Commentary |
|
FY2023 |
Negative |
Net loss year; post-acquisition integration drag |
|
FY2024 |
~2–3% |
Turning positive; trough year |
|
FY2025 |
~4–5% |
~$290M operating income on ~$14.9B total assets |
|
FY2026 (trajectory) |
Improving materially |
Margin expansion + debt reduction driving improvement |
|
Incremental ROIC (InP capacity) |
~15–20%+ (est.) |
Datacenter segment alone, on new capacity dollars |
The blended ROIC is depressed by the 2022 acquisition goodwill (~$8B+). The critical insight is incremental ROIC on new InP capacity — here the returns are genuinely high and well above cost of capital. The NVIDIA $2B investment for InP expansion is structurally capital-light from Coherent's perspective, improving incremental returns further.
Is management converting retained earnings into shareholder value? Partially yes. $437M in debt repaid in FY25. The NVIDIA investment de-risks the InP capex program. SBC remains a watchpoint. Full ROIC recovery requires goodwill amortization roll-off or sufficient operating income growth to overwhelm the asset base.
E. VARIANT PERCEPTION — THE MOST IMPORTANT SECTION
What Consensus Believes
• Coherent is a high-quality AI infrastructure compounder with secular tailwinds
• NVIDIA deal validates the moat
• Revenue grows from ~$7B (FY26) to ~$9.5B (FY27) per consensus
• Margins continue expanding toward 42%+ gross margin; debt manageable and declining
Bull Variant — What Consensus May Be Underestimating
The market may still materially underestimate the CPO and Optical Circuit Switching (OCS) opportunity. These are not incremental product lines — they are architectural shifts in data center design worth tens of billions in TAM. If Coherent wins CPO qualification at NVIDIA's next GPU platform (Rubin/Feynman), the revenue trajectory diverges sharply from current consensus. Management has explicitly guided that FY2027 growth will exceed FY2026 — despite consensus already projecting nearly 34% YoY revenue growth in FY2027. This is a rare instance of management being more bullish than an already-aggressive base case.
Bear Variant — What Consensus May Be Overestimating
The market may be confusing demand visibility with earnings quality. Book-to-bill of 4x is real, but GAAP net income was just $49M on $5.8B of revenue in FY25 (0.8% margin). The non-GAAP adjustments — primarily amortization from the 2022 acquisition — add back hundreds of millions annually. If AI capex slows in 2027 (hyperscaler digestion cycle), orders placed today may get pushed, creating a demand air pocket that the current multiple cannot absorb.
The real mispricing, if any, is at the laser/chip layer of the stack — not the module layer. Framing Coherent as a laser company rather than a transceiver company changes the valuation story meaningfully and justifies a premium over pure module assemblers.
F. INVESTMENT KILL CRITERIA
|
Kill Criterion |
Status |
Assessment |
|
Deteriorating moat |
CLEAR |
InP moat strengthening; module layer watchpoint |
|
ROIC below cost of capital |
WATCH |
Blended ROIC improving; incremental ROIC strong |
|
Secular decline |
CLEAR |
Secular tailwinds among strongest in public markets |
|
Excessive leverage |
WATCH |
$3.7B debt; manageable but not comfortable |
|
Value-destructive M&A |
CLEAR |
Anderson divesting non-core; no new large acquisitions |
|
Misaligned incentives |
WATCH |
SBC elevated; monitoring required |
|
Extreme valuation optimism |
WATCH |
60x non-GAAP; no room for execution error |
Kill criteria not triggered. Watch items: leverage trajectory, SBC dilution, and valuation multiple — all require monitoring rather than immediate action.
G. FINANCIAL DEEP DIVE
Revenue Trajectory
|
Fiscal Year |
Revenue |
YoY Growth |
Commentary |
|
FY2023 |
$5.16B |
— |
Post-acquisition trough approaching |
|
FY2024 |
$4.71B |
-8.8% |
Optical market downturn; integration trough |
|
FY2025 |
$5.81B |
+23.4% |
Record; AI datacenter demand surge begins |
|
LTM (Mar 2026) |
$6.60B |
+18% |
Continued acceleration |
|
FY2026 Full Year (est.) |
~$7.0B |
~20% |
Q4 FY26 guidance mid-point |
Profitability & Margin Expansion
|
Metric |
FY2025 |
Q3 FY26 |
Long-Term Target |
|
GAAP Gross Margin |
35.2% |
37.7% |
42%+ |
|
Non-GAAP Gross Margin |
37.9% |
39.6% |
42%+ |
|
GAAP Net Income |
$49M (0.8% margin) |
$0.97/share (Q) |
Improving |
|
Non-GAAP EPS |
$3.53 (+191% YoY) |
$1.41/quarter |
$7–8 FY27E (est.) |
|
Debt Reduction (FY25) |
$437M repaid |
Continuing |
Target <$2B |
Balance Sheet Snapshot
|
Item |
Jun 2025 |
Mar 2026 (post-NVIDIA) |
|
Total Debt |
$3.69B |
$3.7B (approx., pre-paydown) |
|
Cash |
$909M |
~$3B+ (NVIDIA $2B received) |
|
Total Equity |
$6.0B |
$11.0B (NVIDIA investment) |
|
Total Assets |
$14.9B |
$17.3B |
|
Net Debt (approx.) |
~$2.8B |
Improving significantly |
H. CAPITAL ALLOCATION EXCELLENCE
|
Area |
Rating |
Commentary |
|
Acquisitions |
Mixed |
2022 Coherent deal expensive; synergies achieved; debt burden significant |
|
Divestitures |
Strong |
A&D business + Munich operations sold; strategic focus improving |
|
Debt Reduction |
Strong |
$437M repaid FY25; trajectory ahead of initial pace |
|
Buybacks |
None |
Correct given debt load; preserve cash for debt and capex |
|
Dividends |
None |
Correct for growth stage |
|
Organic Investment (InP 6-inch) |
Excellent |
High-conviction reinvestment in the core moat |
|
External Financing (NVIDIA) |
Excellent |
$2B non-dilutive capital for strategic capex — textbook smart financing |
The NVIDIA deal is a capital allocation masterstroke: customer funds your capacity expansion in exchange for supply priority. It reduces equity dilution, validates the technology, and de-risks the capex program simultaneously. Comparable to how TSMC's anchor customers commit to node capacity.
I & J. MANAGEMENT, GOVERNANCE & CREDIBILITY SCORECARD
Leadership
CEO: Jim Anderson — Former AMD (President, Computing & Graphics). Focused agenda: divest non-core, expand margins, reduce debt, deepen photonics moat. NVIDIA partnership is the capstone of this thesis.
CFO: Sherri Luther — Accurate guidance track record; FY25 358 bps margin expansion delivered as committed. Insider holdings grew meaningfully since Sept 2025 (285 → 12,610 shares); small scheduled sale is liquidity event, not conviction loss.
Credibility Scorecard
|
Metric |
Score |
Commentary |
|
Guidance Accuracy |
High |
FY25 margin expansion delivered; non-GAAP EPS +191% delivered |
|
Strategic Execution |
High |
Divestitures completed; NVIDIA deal closed; S&P 500 inclusion |
|
Communication Quality |
High |
Transparent GAAP vs non-GAAP; clear long-term targets |
|
Capital Allocation |
Improving |
Debt reduction ahead of pace; NVIDIA deal structure excellent |
|
Overall Credibility Score |
7.5/10 |
Anderson executing well; inherited acquisition complexity is the drag |
K. INDUSTRY STRUCTURE & COMPETITIVE DYNAMICS
Porter's Five Forces
|
Force |
Intensity |
Commentary |
|
Competitive Rivalry |
High |
Lumentum, Innolight, Cisco-Acacia, Marvell in adjacent layers |
|
Supplier Power |
Low-Medium |
Coherent IS the supplier; InP is scarce and Coherent controls the bottleneck |
|
Buyer Power |
High |
Hyperscalers + NVIDIA have enormous purchasing leverage |
|
Threat of Substitutes |
Medium |
CPO could reduce pluggable TAM; Coherent positioning for CPO transition |
|
New Entrants |
Low |
Capital intensity + 20-year InP learning curve is nearly impenetrable |
Key Industry Numbers
• 800G/1.6T market projected to grow at ~40% CAGR through 2033 (analyst consensus)
• 1.6T data rates specifically expected to grow at ~200% CAGR to reach $21B market
• Coherent's Networking segment grew 49% YoY in FY2025
• Company's own 2030 SAM estimate for Datacenter & Communications: $32B
M & N. EARNINGS ANALYSIS & CALL INTELLIGENCE
|
Quarter |
Revenue |
GAAP/Non-GAAP GM |
Non-GAAP EPS |
YoY EPS Δ |
|
Q4 FY25 |
$1.53B (record) |
35.7% / 38.1% |
$1.00 |
— |
|
Q1 FY26 |
$1.58B |
~38.5% / ~39%+ |
~$1.19 (est.) |
~26% |
|
Q2 FY26 |
$1.69B |
36.9% / 39.0% |
$1.29 |
+35% |
|
Q3 FY26 |
$1.81B |
37.7% / 39.6% |
$1.41 |
+55% |
What Management Is Signaling Beyond the Numbers
• Book-to-bill >4x in Datacenter: Most of CY2026 booked; CY2027 filling quickly. Strongest visibility in company history.
• 1.6T margins confirmed higher than 800G: Higher ASPs early in lifecycle + 6-inch InP cost advantages. This is a product-level margin confirmation, not an assumption.
• CPO is not a 2029 story: Every call since Q1 FY26 mentions CPO and OCS alongside 1.6T — signals a 2027–2028 revenue entry point.
• 4x InP capacity by end-2027: Management is investing ahead of demand. This is high-conviction capital allocation.
Q. RISK ASSESSMENT
|
Risk |
Probability |
Impact |
Severity |
|
AI capex cycle slowdown / digestion |
Medium |
Very High |
🔴 HIGH — single biggest risk |
|
Chinese vendor module price compression |
High |
Medium |
🟡 MODERATE — module layer less protected |
|
CPO transition execution miss |
Medium |
High |
🔴 HIGH — thesis-level risk |
|
Balance sheet / debt covenant stress |
Low-Medium |
High |
🟡 MODERATE — NVIDIA investment materially improves |
|
Customer concentration |
High |
Medium |
🟡 MODERATE — NVIDIA + 5 hyperscalers dominate |
|
Geopolitical risk (China ~11% rev) |
Medium |
Medium |
🟡 MODERATE — monitor tariff / export dynamics |
|
GAAP/Non-GAAP earnings gap |
Ongoing |
Low |
🟢 LOW — amortization known; disclosures transparent |
|
SBC dilution |
Ongoing |
Low |
🟢 LOW/WATCH — 2–3% of revenue annually |
U. PEER ANALYSIS
|
Company |
Rev (LTM) |
Rev Growth |
ROIC Profile |
EV/Rev |
Notes |
|
Coherent (COHR) |
$6.6B |
~18% |
Low→improving |
~10x |
InP moat + NVIDIA deal |
|
Lumentum (LITE) |
~$2.2B |
+40%+ |
Medium |
~8–9x |
800G strength; no InP integration |
|
Marvell (MRVL) |
~$8B |
+60%+ |
Med-High |
~15x |
Custom silicon; strong CPO play |
|
Cisco / Acacia |
Large |
Low |
High |
Premium |
Diversified; less pure-play |
Coherent trades at a discount to Marvell on EV/Revenue despite more direct exposure to the transceiver layer. The discount is partially justified by lower current ROIC and residual debt. If margins reach 42%+ and debt falls below $2B, the gap narrows sharply.
V, W & X. VALUATION & EXPECTED RETURN ANALYSIS
Relative Valuation (at ~$350/share)
|
Metric |
Value |
Commentary |
|
Non-GAAP P/E (FY26E ~$5.50–6.00) |
~58–64x |
Premium; growth rate partially justifies |
|
EV / Revenue (FY26E ~$7B) |
~10x |
In-line with AI infra peers |
|
EV / Non-GAAP EBITDA |
~35–40x |
Elevated; compressing with margin expansion |
|
Non-GAAP FCF Yield |
~2–3% |
Low but improving rapidly |
Bull Case — What Could Make This Stock Double or Triple
• CPO platform win at NVIDIA Rubin/Feynman: If InP lasers designed into co-packaged optics at scale, TAM expands from ~$20B to $50B+
• Optical Circuit Switching (OCS) becomes standard datacenter architecture: Coherent is one of 2–3 companies globally capable. Management's own 2030 SAM estimate is $32B.
• InP supply bottleneck becomes a permanent structural toll: TSMC-like economics — high margins, long-term supply agreements, pricing power that compounds for years.
• Silicon Carbide strategic monetization: SiC JV with Denso and MELCO could be monetized separately, unlocking hidden balance sheet value.
Bear Case — Five Strongest Arguments Against
• GAAP earnings are thin: 0.8% GAAP net margin on $5.8B revenue. Non-GAAP adjustments are intellectually defensible but real cash was spent. $3.7B debt is the cash cost of the 2022 acquisition sitting on the balance sheet.
• Module commoditization is coming: Chinese vendors improving quality and cutting module prices. The InP laser moat is real; the module assembly moat is not.
• AI capex is the only thesis: Datacenter & Communications is 75% of revenue. Any hyperscaler capex slowdown hits earnings estimates severely.
• Customer concentration is extreme: NVIDIA + 5 hyperscalers dominate. NVIDIA's $2B investment is confidence and dependency simultaneously.
• Stock prices perfection at 60x non-GAAP: Multiple expansion requires EPS to grow faster than stock from here. Any delay collapses the multiple faster than earnings can recover.
AA. BASE RATE ANALYSIS — HISTORICAL ANALOGUES
|
Analogue |
Similarity |
Outcome |
Key Lesson |
|
TSMC (2010–2015) |
Vertically integrated chip mfg; capacity moat |
Multi-bagger |
Moat compounded when technology lead held; capex self-reinforcing |
|
Finisar (pre-II-VI) |
Optical transceiver mfg |
Boom-bust cycles |
Modules alone are cyclical; laser integration is the moat |
|
Lam Research (2015–2020) |
Semicon equipment; capex-cycle dependent |
Multi-bagger during buildout |
Works well until capex digestion; size matters for cycle resilience |
|
Viavi Solutions |
Optical components post-JDS Uniphase |
Moderate; multiple compression |
Commodity optics without vertical integration = weak moat |
The TSMC analogue is most instructive. TSMC's moat came from process node leadership + willingness to out-invest everyone else in manufacturing. Coherent's 6-inch InP bet is a structurally similar move — own the manufacturing bottleneck, let customers come to you.
AD. RED FLAG CHECKLIST
|
Flag |
Status |
Detail |
|
Large non-GAAP adjustments |
🟡 Watch |
Amortization add-back from 2022 acquisition — large but standard post-M&A |
|
Goodwill / intangibles |
🟡 Watch |
~$8B+ from Coherent Inc. acquisition; impairment risk in downturn |
|
Receivables / cash conversion |
🟢 Clear |
Revenue growth matched by cash flow improvement |
|
Related-party (NVIDIA as customer + investor) |
🟡 Watch |
Appropriate disclosure; non-exclusive agreement is the right structure |
|
SBC levels (2–3% of revenue) |
🟡 Watch |
Elevated but not exceptional for semiconductor companies |
|
Debt covenants |
🟢 Clear |
SOFR + 1.50% on term loans; trajectory well within bounds |
|
Insider selling |
🟢 Clear |
CFO small scheduled sale vs. dramatically expanded holding; not concerning |
|
China revenue exposure (~11%) |
🟡 Watch |
Geopolitical and tariff risk; monitor export restriction dynamics |
AE. INVESTMENT SCORECARD
|
Category |
Weight |
Score |
Commentary |
|
Economic Moat |
20% |
7.5 / 10 |
InP laser moat genuine; module layer contestable |
|
Financial Strength |
15% |
6.0 / 10 |
Debt declining; GAAP thin; non-GAAP robust |
|
Capital Allocation |
15% |
8.0 / 10 |
NVIDIA deal structure excellent; divestitures correct |
|
Management |
10% |
7.5 / 10 |
Anderson executing well; inherited acquisition complexity |
|
Growth |
15% |
9.0 / 10 |
20–40% growth; CPO optionality; best visibility ever |
|
Valuation |
15% |
5.5 / 10 |
60x non-GAAP P/E; full price; no margin for error |
|
Risk |
10% |
6.0 / 10 |
AI cycle + Chinese competition + customer concentration |
|
TOTAL WEIGHTED SCORE |
100% |
72.3 |
AG. FINAL VERDICT
Investment Decision
|
Category |
Rating |
Commentary |
|
Economic Moat |
7.5 / 10 |
InP process moat is real; module layer more contestable |
|
Management |
7.5 / 10 |
Executing well; inherited complexity managed cleanly |
|
Capital Allocation |
8.0 / 10 |
NVIDIA deal is a capital allocation masterstroke |
|
Financial Strength |
6.0 / 10 |
Improving fast; $3.7B debt still a live risk |
|
Growth |
9.0 / 10 |
Best near-term visibility in the sector |
|
Valuation |
5.5 / 10 |
Full price; 60x non-GAAP; no margin for error |
|
Risk |
6.0 / 10 |
AI cycle + Chinese competition + concentration |
Bull Case
• The 6-inch InP manufacturing moat is a genuine, 20-year process advantage that no competitor can replicate on a 3–5 year timeline — and it is getting stronger, not weaker.
• The NVIDIA $2B strategic investment + multi-year purchase commitment is structurally equivalent to a hyperscaler signing a 5-year supply contract. Demand visibility is the best in the company's history.
• The 800G to 1.6T to CPO transition is not a threat — it is the next chapter. Coherent's multi-architecture 1.6T approach + NVIDIA CPO partnership positions it to be the enabling infrastructure for the entire next architectural generation.
Bear Case
• Stock prices perfection at ~60x non-GAAP earnings. Any miss — AI capex deceleration, margin disappointment, CPO delay — will be punished severely and quickly.
• The GAAP earnings story (0.8% net margin in FY25) is a reminder that $7–8B+ of acquisition goodwill sits on the balance sheet. The non-GAAP adjustments are defensible but real cash was spent.
• Chinese vendors compete hard on modules and hyperscalers have leverage to keep module pricing in check. The margin story depends on Coherent migrating up the value stack faster than Chinese vendors migrate up the quality stack.
What Would Change The Thesis
• Meaningful reduction in hyperscaler capex guidance from any major cloud provider
• Evidence that CPO transitions are displacing pluggable revenues faster than Coherent wins CPO sockets
• Debt covenant stress or FCF below $800M in FY26
Most Important Metric to Monitor: 1.6T gross margin trajectory — if 1.6T margins are genuinely higher than 800G (as management has guided), the compounding effect on overall gross margin justifies the valuation. If 1.6T margins disappoint, the entire premium collapses.
One-Sentence Investment Thesis: Coherent is the only vertically integrated photonics manufacturer with a process-level moat in indium phosphide lasers, a NVIDIA-endorsed supply position, and a generational upgrade cycle from 800G to 1.6T to co-packaged optics that compounds its competitive advantage with every dollar of new capacity.
CII Research | AI Infrastructure Series | June 2026 | Manu | For informational purposes only — not SEBI-registered investment advice. Do your own due diligence.
⬇ Full research report (.docx) ⬇ Investment committee memo (.docx)
Sources: Coherent Corp. quarterly filings and earnings materials, NVIDIA partnership disclosures, company investor presentations (as of June 2026). Disclaimer: personal research for educational purposes; not investment advice.