Coherent Corp

• Vertical integration: InP substrates → laser diodes → transceivers — unique in the industry
• NVDA $2B investment validates strategic importance for AI GPU cluster optics
• Book-to-bill >4x in datacom — multi-year demand visibility
• Diversified across materials, components, and systems — less single-market risk than LITE
• Highest tech_moat_score (6.7) in our sensor stack

The physical constraint is InP laser (EML/CW) output — the photon source every 800G/1.6T pluggable and every silicon-photonics/CPO module needs. COHR is the only player publicly scaling to 6-inch InP at production yields; the 6-inch platform delivers ~4× die-per-wafer vs 3-inch with higher yields, and the company is on a path to ~4× total InP output by 2027. A competitor cannot buy this position — it must build a 6-inch InP fab and close the yield gap, which is a multi-year, multi-hundred-million-dollar effort. Book-to-bill >4× in datacom (Q3 FY26 disclosure, carried forward) means demand is rationing capacity.

The scarce asset is qualified indium-phosphide laser output (EML / CW / photodiodes) that every 800G/1.6T pluggable and every silicon-photonics / CPO module still needs. Coherent's own 10-K names InP "a foundational technology for next-generation AI optical interconnects" and says it is expanding 6-inch InP capacity in the U.S. and Europe, including Sherman, TX, "to address our increased customer demand and industry-wide shortage." Assembly/test is not the gate; crystal + epi + yield is.

Special $100M-target PSU grant, 8-K filed 2026-08-31 (event date 2026-08-27). Compensation Committee awards, 100% PSUs, outside the regular program, to retain the team through "technology investment, capacity expansion, customer engagement and operational execution" for next-gen AI/compute:

Hurdles (60-day average price and rTSR above median vs S&P Composite 1500 Electronic Equipment, Instruments & Components; no interpolation except on CoC):

No vest until the four-year performance period ends; +1 year holding → shares untradable until 2031. Voluntary quit / for-cause = full forfeit. The 8-K also states: since Anderson became CEO (June 2024) Coherent delivered >300% TSR and record FY2026 revenue.

This is a retention / revealed-preference catalyst, not a demand 8-K. It does not change InP physics. It does tell you the board thinks the next four years of capacity build are the whole game — and that the team is flight-risk without a special grant.

Grant-implied starting price (all four hurdles are internally consistent):

• $454.43 / (1.10)^4 = $310.38
• $542.86 / (1.15)^4 = $310.38
• $643.61 / (1.20)^4 = $310.38
• $757.77 / (1.25)^4 = $310.38

So the Committee underwrote a ~$310 reference price on 2026-08-27.

Read: dilution is small (sub-0.4% even at 200%). Economic meaning is not share count — it is 12% of last year's earnings set aside as a four-year retention coupon on the people running the InP ramp. Compare to the real cost of the moat: R&D $723M + capex $1,103M = $1,826M against Datacenter & Communications segment profit $1,330M → 1.37× one year of DC profit spent to defend/extend the bottleneck. The PSU is a rounding error next to that capex; the capex is the tell.

FY2026 GAAP P&L (10-K MD&A + companyfacts):

Segment (Note 20, $000):

Industrial −10% is mostly the aerospace & defense sale (2025-09-02) and Munich sale (2026-01-30) — not a second growth engine dying in place. After those divestitures the company is a 74% AI-optics name.

Cash (the 08-16 note did not dollarize this):

MD&A (verbatim substance): OCF fell $634M → $80M "primarily driven by a significant increase in inventory levels to support higher revenue growth," partially offset by higher payables and higher earnings. They also prepaid $1,723M of existing debt and spent $1,103M on PP&E. Liquidity was backfilled by the NVIDIA $2B equity, not by operations.

Cheat-sheet prior: PA Layers 1–8 mostly 6+. Optical interconnect is Layer 6. DB already had STRONG_PROXY(7) with concentration=3; that 3 is too high for 37.5% GAAP GM and 32% top-2.

• product_service: photonic and compound semiconductor solutions: lasers, optical transceivers, optics, silicon carbide, optical materials. Source: 10-K Business.
• technical function: generate, modulate, transmit, detect light; enables high-speed datacom/telecom, precision industrial processing, sensing, semiconductor manufacturing. Source: 10-K Business.
• why it is hard: requires epitaxial growth, precision optics, materials science, packaging, vertical integration, IP, and customer design qualification. Source: 10-K Business/Risk Factors.
• substitutes: other transceiver/laser vendors, silicon photonics, electrical interconnects for short reach, in-house hyperscaler optics. Source: 10-K Risk Factors.
• pricing power: MED — differentiated photonics/compound semi, but competition and large-customer pricing pressure. Source: 10-K Risk Factors.

Coherent’s products fall into three groups [1]:

1. Networking

• Datacom optical transceivers, including 800G and emerging 1.6T pluggable modules for AI clusters.
• Coherent optical transceivers and components for telecom.
• VCSELs, photodiodes, pump lasers, and optical subassemblies.

1. Materials

• Silicon carbide substrates for power electronics.
• Indium phosphide wafers.
• Gallium arsenide wafers.
• Other engineered materials used in optoelectronics and RF devices.

1. Lasers

• CO₂ lasers, fiber lasers, diode lasers, ultrafast lasers, and excimer lasers.
• Laser systems and beam-delivery components for welding, cutting, marking, and materials processing.

The core technical function is optical-electrical conversion and light generation. For AI datacom, an optical transceiver converts electrical signals from a switch or accelerator into modulated light, transmits it over fiber, and converts it back to electrical at the other end. Short-reach copper can substitute for very short links, but at 800G and beyond, optical links are required for reach, density, and bandwidth.

Silicon carbide substrates are the foundation for power devices used in electric vehicles, industrial power supplies, and energy infrastructure. Lasers are used to cut, weld, drill, mark, or process metals, polymers, semiconductors, and other materials.

Substitutes exist but are not perfectly elastic across the portfolio:

• Direct-attach copper replaces optical links inside a rack or for very short reach, but not for rack-to-rack or campus-scale AI cluster links.
• Silicon photonics and thin-film lithium niobate are alternative material/device platforms, but Coherent also participates in or supplies components into these ecosystems.
• Competing transceivers from Lumentum, Innolight, Broadcom, Cisco, and others.
• Competing laser platforms from IPG Photonics, Trumpf, and MKS Instruments.
• Silicon carbide competitors such as Wolfspeed, Rohm, and II-VI’s own former Compounds business are now part of Coherent; external competitors include numerous substrate producers.

The 10-K repeatedly emphasizes vertical integration from engineered materials through components to finished modules and systems [1]. This is a real barrier: a pure module maker must buy wafers, lasers, and photodiodes from third parties, while Coherent can source many of those internally. Customer qualification cycles in optical transceivers and power electronics are long, and high-reliability lasers require years of applications engineering.

That said, “vertical integration” does not by itself guarantee pricing power. SiC substrate pricing has been under pressure from industry capacity additions, and older-speed optical transceivers commodity-ize quickly.

Pricing power is strongest in leading-edge datacom optical transceivers, where capacity is constrained and customers are racing to deploy AI clusters. Coherent can likely raise prices or hold pricing on 800G/1.6T optical modules without losing the highest-volume AI sockets.

Pricing power is weakest in older-speed transceivers, generic industrial lasers, and merchant silicon carbide substrates, where buyers can dual-source or substitute. Therefore, Coherent’s aggregate pricing power is real but segment-dependent.

Coherent’s product portfolio can be broken into the three segments:

1. Networking

Coherent sells optical components, modules, and subassemblies for:

• Datacom and telecom optical transceivers,
• Pump lasers and optical amplifiers,
• Wavelength-selective switches (“WSS”),
• Reconfigurable optical add-drop multiplexers (“ROADMs”),
• Optical circuit switches,
• Vertical-cavity surface-emitting lasers (“VCSELs”) for 3D sensing,
• Coherent optics for high-speed networking.

This is not a router or switch company. Coherent supplies the photonic engines that go inside network equipment. Its products convert electrical signals to optical signals, route wavelengths, amplify light, and receive signals at the other end.¹

2. Materials

Coherent grows silicon carbide crystals, manufactures SiC wafers and epitaxial wafers, and supplies complex engineered ceramics, high-temperature materials, and optical crystals.¹ SiC is used in power semiconductors for electric vehicles, industrial motor drives, and charging infrastructure. The key technical function is enabling power devices that operate at higher voltage, higher frequency, and higher temperature than silicon devices.

3. Lasers

Coherent sells lasers and laser-based systems:

• Fiber lasers,
• Diode lasers,
• Solid-state lasers,
• CO₂ lasers,
• Excimer lasers,
• Laser systems for cutting, welding, ablation, annealing, and display manufacturing.

The technical function is controlled energy delivery. Laser systems are used to cut sheet metal, weld battery components, anneal semiconductor wafers, and manufacture OLED displays.¹

Coherent has several moats that are visible in its filings and in its product disclosures:

• Vertical integration in exotic materials. It grows its own crystals and produces its own indium phosphide and gallium arsenide semiconductor wafers. This is dirty, capital-intensive, and difficult to duplicate quickly.
• Long customer qualification cycles. Optical transceivers for hyperscale datacenter customers require design-in, qualification, reliability testing, and multi-generational interoperability testing. Once a supplier is qualified, switching creates engineering risk.
• Trade secrets and process know-how. SiC crystal growth yield and optical coating stability are yield-sensitive processes. Coherent states that its manufacturing processes are a competitive asset.¹
• Broad portfolio breadth. Few competitors span datacom transceivers, SiC materials, and industrial laser systems. This breadth gives Coherent multiple demand cycles but also creates complexity.

Substitutes vary by segment:

This table is assembled from industry context and competitor filings; it is not a market-share assertion. The load-bearing market-position claims should be treated as [SINGLE-SOURCE] unless verified against an independent market-research source.

Coherent’s pricing power is mixed:

• In differentiated laser systems and precision optical modules, pricing power is moderate. Customers buy reliability, process know-how, and ramp speed, not just price.
• In standard datacom transceivers, pricing is highly competitive and declines rapidly. Coherent itself discloses in the 10-K that the industry experiences price erosion for optical components.¹
• In SiC materials, pricing power depends on supply/demand. During SiC substrate shortages, pricing was strong; in oversupply periods, pricing weakens. This cyclicality is visible in competitor Wolfspeed filings, though Coherent does not break out SiC pricing in the 10-K.¹

A useful signal on pricing power is gross margin. Coherent’s FY2024 GAAP gross margin of about 30.5% is respectable for a hardware company, but the path to higher gross margin depends on mix improvement toward higher-speed datacom transceivers and higher-margin laser systems.¹

Coherent’s products fall into three broad buckets:

1. Networking components and transceivers

Coherent sells optical transceivers, optical subassemblies, lasers, receivers, and components used in telecom and datacom networks. This includes high-speed pluggable modules for data centers, metro/long-haul optical systems, and internal laser sources.

The technical function is converting electrical signals to modulated optical signals and back. High-speed modules require:

• Indium phosphide lasers or silicon photonics chips.
• Modulation and photodetection.
• Optical coupling and packaging.
• Thermal and electrical integration.

1. Engineered materials

Coherent sells substrates and materials such as silicon carbide, indium phosphide, gallium arsenide, and optical materials. Silicon carbide substrates are used in power electronics and RF devices. Indium phosphide is a foundational material for high-speed optical lasers and detectors.

The technical function is enabling electronic/photonic devices that work at high voltage, high frequency, or high optical efficiency.

1. Lasers and systems

Coherent sells laser sources, beam delivery subsystems, and laser-based machines. Applications include cutting, welding, marking, medical procedures, and scientific research.

The harder-to-replicate parts of Coherent’s business are:

• Crystal growth and epitaxy — producing high-quality SiC or InP wafers requires deep process know-how, low-defect growth, and long yield-learning cycles.
• Laser-to-module integration — the transition from a bare laser chip to a qualified 800G or 1.6T transceiver is not trivial. Customers qualify modules for heat, signal integrity, reliability, and interoperability.
• Qualification cycles — once Coherent is designed into an OEM system, replacing it is slow and risky.

That said, the moat is not absolute. Competitors such as Lumentum, Cisco, Broadcom, and Asian module makers compete in optical transceivers. Silicon carbide substrates face expanding Chinese supply. The bull case should not assume permanent pricing power.

• Silicon photonics — replaces some discrete InP lasers in short-reach optics, though Coherent also participates in silicon photonics.
• VCSELs — substitute for edge-emitting lasers in short-reach data links.
• Co-packaged optics (CPO) — substitutes for discrete pluggable modules by integrating optics closer to switch chips.
• Direct attach copper — substitutes for optical interconnects in very short AI rack-scale links.
• Alternative SiC suppliers — pressure materials pricing.

Pricing power is strongest when Coherent is selling capacity-constrained advanced transceivers or qualified materials during a ramp. It is weakest in mature telecom transceivers and commodity optical components.

The key question: can Coherent raise prices without losing share in AI datacom transceivers? The likely answer from management commentary is that pricing is firm during the 800G/1.6T ramp because supply is constrained and customers prioritize availability. But in a normal cyclical downturn, transceiver pricing has historically declined. Therefore, pricing power is cyclical rather than structural.

Coherent’s product portfolio spans three technical layers:

The Networking segment sells optical building blocks for high-speed communications:

• Pluggable optical modules — 400G, 800G, and emerging 1.6T datacom transceivers for Ethernet switches and AI-cluster interconnects.
• Coherent optical components — lasers, modulators, receivers, and digital signal processing assembled into coherent transceivers for long-haul/metro telecom.
• Optical amplification and switching — erbium-doped fiber amplifiers, Raman amplifiers, wavelength-selective switches, and ROADMs used in carrier networks.
• Optical passives — filters, thin-film coatings, isolators, and passive fiber assembly.

Technical function: convert electrical signals to optical signals, transmit them over fiber, and amplify/route them with minimal loss.

The Lasers segment sells:

• Fiber lasers for cutting, welding, and marking.
• Ultrafast lasers for micro-machining, medical, and scientific use.
• CO2 lasers for processing organic materials, glass, and ceramics.
• Excimer lasers used in microelectronics and refractive surgery.
• Laser systems and subsystems for semiconductor, display, and battery manufacturing.

Technical function: deliver controlled, high-energy light for industrial or medical material processing.

The Materials segment sells engineered materials:

• Silicon carbide substrates — critical for power electronics and emerging 800V EV platforms.
• Optical crystals and ceramics — used for laser gain media and optical assemblies.
• Thermal materials — for high-power electronics cooling.
• Semiconductor materials — including indium phosphide and gallium arsenide substrates or epitaxial wafers.

Technical function: provide the physical substrate or optical medium from which active photonic/power devices are made.

Coherent’s defensibility comes from vertical integration. It controls materials science, device fabrication, packaging, and system-level test. That is more difficult to replicate than a pure module-assembly business. The silicon carbide substrate process, optical-coating expertise, and high-volume fiber-laser manufacturing create engineering and process barriers.

That said, barriers are not infinite. In optical transceivers, competitors such as Lumentum, Broadcom, Cisco, Nokia, and Asian module makers can supply similar form factors. In silicon carbide, Chinese substrate producers have expanded capacity aggressively, creating price pressure.

Coherent has selective pricing power, not broad pricing power:

• Stronger pricing power: silicon carbide substrates, differentiated high-speed optical components, specialized lasers where customers depend on process qualification.
• Weaker pricing power: standard optical transceivers, mature industrial lasers, and commodity materials where competition is intense.

The company’s own competitor, Lumentum, describes intense price competition in optical markets. That disclosure contradicts any assumption that Coherent can raise prices broadly without share loss.

Source: Lumentum Holdings Inc. FY2024 Form 10-K, CIK 0001632337, filed August 2024, accessed 2026-04-26.

URL: sec.gov

• Question: Latest 2026-07-17 signal: RED alert; funnel_tier=PORTFOLIO; flow_score=4.0; reason=pre-mkt -7.5%
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• Question: PSU 8-K 08-31: $100M target retention PSUs (CEO $50M) with 10-25% 4y CAGR hurdles from implied $310.38 ($454.43/$542.86/$643.61/$757.77) + rTSR>median; vest 2030 +1y hold. FY26 10-K: rev $7,118M +22.5%, GAAP GM 37.5%, NI $805M; DC $5,275M 74.1% mix +40.5% YoY, segment profit $1,330M (25.2%); Industrial $1,844M -10% on A&D/Munich divestitures. OCF $80M vs NI $805M (OCF/NI 9.9%) on +$1,143M inventory, DIO 212d; FCF -$1,023M after $1,103M capex. Customers 20%+12%=32% both in DC. NVIDIA multi-year multi-billion purchase + $2B equity 03-02. Taxonomy Layer7 perception -> Layer6 connectivity. Scores: ts 8 / ec 7 / ttm 7 (from 6.0/9.0/4.0). Toll-road 7 STRONG_PROXY -> 6 STRONG_PROXY (bypass 2 / conc 2 / hyper 2). Verdict INTACT.
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• Question: Latest 2026-07-17 signal: RED alert; funnel_tier=PORTFOLIO; flow_score=4.0; reason=pre-mkt -7.5%
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• Question: Latest 2026-07-17 signal: RED alert; funnel_tier=PORTFOLIO; flow_score=4.0; reason=pre-mkt -7.5%
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• Question: Q4 FY26 beat-and-raise: rev $2.05B +34% YoY, non-GAAP GM 40.2%, FY26 non-GAAP EPS $5.61 (+59%). Datacom & Comm 79% mix +58.6% YoY; Industrial -15.7%. 6-inch InP (4.0x die/wafer) ramp to ~4x output by 2027. Scores: ts 8 / ec 8 / ttm 8. Verdict STRENGTHENED.
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• Question: Latest 2026-07-17 signal: RED alert; funnel_tier=PORTFOLIO; flow_score=4.0; reason=pre-mkt -7.5%
• Why it matters: Weekend sprint bottleneck capture
• Source needed: weekend_learning_cycle_v2

• [ ] What is the exact LiDAR-specific revenue within the $400-450M industrial/laser segment?
• [ ] When does the InP capacity doubling reach full production?
• [ ] Is the NVDA partnership exclusive or will COHR supply other GPU/switch vendors?

COHR is the higher-quality version of the LITE thesis. Broader portfolio, vertical integration, NVDA-validated, highest moat score in our sensor stack (6.7). But the same caveat applies: this is an AI photonics play, not a pure LiDAR bet. Buy COHR for the photonics backbone of AI infrastructure. The LiDAR/sensing segment (~$400M/quarter) is a free option on the sensor buildout.

STRENGTHENED — Q4 FY2026 is a clean beat-and-raise (revenue $2.05B +34% YoY, non-GAAP GM 40.2%, FY27 guide implying ~+29% revenue / +39% EPS), and it confirms the moat has widened: the 6-inch InP transition (4× die/wafer) is a barrier competitors must fund years in advance. The one real caveat is the death of the non-datacom diversification backstop — COHR is now a high-beta, levered AI-optics pure-play, not a diversified photonics compounder.