Coherent vs Lumentum Deep Dive: Coherent Owns the Stack, Lumentum Owns the Scarcity
As of September 22, 2026, Coherent and Lumentum have become 2 of the most strategically important merchant suppliers to the AI optical-connectivity ecosystem. Both companies are participating in the same structural transition: AI clusters are moving from 400G and 800G pluggable optics toward 1.6T, 3.2T, near-packaged optics, co-packaged optics, optical circuit switching and eventually optical scale-up fabrics. Both have also received $2 billion strategic investments from NVIDIA alongside non-exclusive, multi-year purchase commitments. The similarity ends there. Coherent is a vertically integrated, materials-to-systems photonics platform with approximately $7.1 billion of fiscal 2026 revenue and meaningful industrial diversification. Lumentum is a smaller, more concentrated optical specialist whose scarce indium-phosphide laser capacity, cloud-transceiver portfolio and first-mover position in merchant optical circuit switches are producing materially higher incremental margins. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm?utm_source=openai))
The central investment distinction is durability versus torque. Coherent has the broader technology stack, greater manufacturing redundancy, a stronger position in complete transceivers and more ways to monetize the migration from pluggables to integrated optics. Lumentum has the cleaner exposure to the industry’s tightest bottlenecks, particularly 200G-per-lane EMLs, high-power continuous-wave lasers and OCS systems. Lumentum is therefore the sharper near-term beneficiary of optical scarcity, while Coherent is the more diversified long-duration platform. The correct comparison is not which company has more AI exposure; both have substantial exposure. It is which layers of the optical stack will retain pricing power once current shortages ease, Chinese capacity expands and hyperscalers begin internalizing more of the system architecture.
The Business Models Are Converging, but They Remain Structurally Different
Coherent develops and manufactures lasers, optical transceivers, semiconductor devices, photonic integrated circuits, optical switches, engineered materials, thermal-management products and complete laser systems. Its fiscal 2026 reporting structure consists of Datacenter & Communications and Industrial. Datacenter & Communications generated approximately $5.27 billion, or 74% of fiscal 2026 revenue, while Industrial generated approximately $1.84 billion. By fiscal Q4 2026, Datacenter & Communications had risen to 79% of company revenue, reflecting rapid growth in AI-related transceivers, optical components, DCI products and communications infrastructure. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm))
The breadth of Coherent’s revenue base is a consequence of its history. The former II-VI combined compound-semiconductor materials, optical components and engineered materials with the legacy Coherent laser franchise. The result is a company that can grow indium-phosphide and gallium-arsenide wafers, fabricate lasers and photodiodes, manufacture passive optical components, produce photonic integrated circuits and assemble those technologies into modules or systems. In Industrial, Coherent sells excimer, solid-state and CO2 lasers, materials-processing systems, laser optics, thermoelectric components, engineered ceramics and materials based on ZnSe, ZnS, GaAs, InP, GaSb and SiC. The company reaches semiconductor-equipment manufacturers, display-equipment vendors, industrial automation customers, medical and scientific-instrumentation suppliers, and communications customers through both direct sales and distribution. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm))
Coherent’s business model is therefore based on monetizing intellectual property and manufacturing assets at several stages of the value chain. It can sell an InP laser chip to a module vendor, integrate that laser into its own transceiver, provide a coherent transport subsystem to a network-equipment manufacturer, or supply a complete optical system. The economic benefit is not merely additional revenue per application. Vertical integration allows the company to allocate scarce components internally, optimize products across component and module boundaries, capture more of the bill of materials and lower costs as manufacturing moves to larger wafers. The economic cost is organizational complexity, elevated capital intensity and exposure to end markets that do not share the growth or margin characteristics of AI optics.
Lumentum is more concentrated. It reports revenue by Components and Systems rather than by end market. Components generated $2.01 billion, or 66.5% of fiscal 2026 revenue, and Systems generated $1.01 billion, or 33.5%. Components include EMLs, continuous-wave lasers, high-power laser assemblies, coherent optical components, pump lasers, wavelength-management devices and 3D-sensing emitters. Systems include cloud transceivers, optical circuit switches, optical transport subsystems and industrial lasers. Fiscal 2026 revenue reached $3.01 billion, up 83%, with Components increasing 80% and Systems increasing 91%. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1633978/000162828026057358/lite-20260627.htm))
Lumentum’s operating model is narrower but not asset-light. It fabricates critical optical semiconductors internally, particularly in Japan, while relying on contract manufacturers in Thailand, Taiwan, Malaysia and the Philippines for selected assembly and manufacturing activities. The company expanded through Oclaro, NeoPhotonics, Cloud Light and IPG Photonics’ telecom-transmission product lines. These acquisitions added coherent components, advanced lasers, cloud transceivers and manufacturing expertise. Cloud Light was particularly important because it moved Lumentum further downstream from component supply into complete hyperscale transceiver modules. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1633978/000162828026057358/lite-20260627.htm))
Lumentum’s fiscal Q4 2026 results illustrate the earnings power of a concentrated portfolio during an optical shortage. Quarterly revenue reached approximately $1.01 billion, non-GAAP gross margin exceeded 50% and non-GAAP operating margin reached 36.6%. Those margins were achieved well below the company’s previously contemplated $2 billion quarterly revenue level, reflecting favorable product mix, improved factory utilization, pricing actions and scarcity in high-performance lasers. Coherent’s fiscal Q4 non-GAAP gross margin was 40.2%. The gap does not imply that all of Lumentum’s products are structurally superior; it reflects a more concentrated mix of bottleneck components, while Coherent carries lower-margin transceiver assembly and a broader industrial portfolio. ([investor.lumentum.com](https://investor.lumentum.com/financial-news-releases/news-details/2026/Lumentum-Announces-Fourth-Quarter-and-Full-Fiscal-Year-2026-Results/default.aspx?utm_source=openai))
The Product Portfolios Reveal Different Ways to Capture AI Optical Spending
Coherent’s principal AI products are 800G and 1.6T optical transceivers, coherent ZR and ZR+ modules, EML and CW laser chips, photodiodes, VCSELs, optical circuit switches, silicon-photonics products, CPO and NPO components, optical fibers, isolators, thermal solutions and integrated optical subsystems. The company supports Ethernet, InfiniBand, NVIDIA NVLink and other architectures, making its portfolio largely protocol-agnostic. That is strategically important because the precise division of AI networking between Ethernet, proprietary scale-up fabrics and alternative interconnect standards remains unsettled. Coherent can participate without having to predict a single winning protocol. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm))
Lumentum’s strongest positions are concentrated in high-speed optical sources and the systems built around them. Its EMLs combine a laser and modulator and are favored where signal integrity and reach requirements are demanding. Its CW lasers supply external light to silicon-photonics engines, while its ultra-high-power lasers are required for CPO and NPO architectures in which one laser source can feed multiple optical channels. Lumentum also supplies coherent transmission components, ROADMs, wavelength-selective switching, pump lasers, amplifiers, cloud transceivers and optical circuit switches. Industrial products include ultrafast and fiber lasers used in semiconductor processing, PCB drilling, EV-battery production and precision manufacturing. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1633978/000162828026057358/lite-20260627.htm))
The companies overlap most directly in transceivers, EMLs, CW lasers, coherent components, OCS, CPO and selected industrial lasers. Coherent has the stronger breadth and the larger complete-transceiver franchise. Lumentum appears stronger in merchant high-performance EMLs and currently has the more advanced commercial OCS ramp. Coherent can capture more total system content when it wins a complete module. Lumentum can generate superior margins when customers require a laser whose performance is difficult to reproduce and whose failure would depress yields across an expensive transceiver assembly.
This distinction matters as 1.6T architectures evolve. At 800G, EML-based designs have been a major solution. At 1.6T, silicon photonics and CW lasers gain share because optical modulation can be integrated into the silicon-photonics engine. That potentially reduces EML content per transceiver even while total laser demand rises. Lumentum expects EML unit volumes to continue growing, but it also acknowledges that EML share may decline during the 1.6T transition before potentially recovering at 3.2T, where signal-integrity challenges again favor integrated laser-modulator solutions. Coherent is relatively insulated from this debate because it sells EMLs, CW lasers, silicon-photonics components and complete modules. Lumentum is managing the transition by allocating more fab output toward CW lasers and shrinking CW die size to improve margins. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Customers, End Customers and Concentration Risk
Both companies sell directly to cloud operators, AI infrastructure providers, networking-equipment manufacturers, optical-module companies, telecom-equipment vendors and industrial OEMs. Their ultimate end customers include hyperscale cloud platforms, AI model developers, neocloud operators, telecommunications carriers, semiconductor manufacturers, display manufacturers and industrial-production companies. The identity of the largest customers is not fully disclosed, and shipment locations can be misleading because products are frequently delivered to contract manufacturers rather than to the headquarters of the ultimate customer.
Coherent’s customer concentration is significant but lower than Lumentum’s. In fiscal 2026, Coherent’s largest customer represented 20% of consolidated revenue and its second-largest represented 12%, with both relationships concentrated primarily in Datacenter & Communications. The top 2 therefore accounted for approximately 32% of revenue. The company’s direct customer concentration understates ultimate platform concentration because several module and network-equipment customers may ultimately serve the same hyperscaler or accelerator ecosystem. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm))
Lumentum’s largest customer represented 26.6% of fiscal 2026 revenue and its second-largest represented 15%, taking the combined share to 41.6%. That concentration has increased as cloud transceivers, high-speed lasers and OCS have ramped. The associated economics are currently attractive, but the dependence is material. A hyperscaler can change module suppliers, alter optical architecture, dual-source a laser, internalize OCS production or negotiate lower pricing once supply catches up. The risk is amplified because Lumentum’s largest opportunities are linked to a small number of very large AI buildouts rather than thousands of independent customers. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1633978/000162828026057358/lite-20260627.htm))
NVIDIA is the only major relationship that both companies have publicly identified at exceptional strategic scale. On March 2, 2026, NVIDIA agreed to invest $2 billion in each company and entered non-exclusive, multi-year arrangements involving multibillion-dollar purchase commitments and future capacity rights. Coherent’s arrangement covers advanced lasers and optical-networking products, while Lumentum’s is focused on advanced laser components and collaborative optical development. The agreements validate both companies’ technology and improve capacity-planning visibility, but they are not exclusivity contracts. NVIDIA intentionally preserved access to multiple sources, reducing supplier-specific risk and maintaining negotiating leverage. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1633978/000119312526085412/d41019dex991.htm?utm_source=openai))
The broader customer environment remains favorable. NVIDIA reported another record networking quarter in fiscal Q2 2027 and said production shipments of Vera Rubin had begun, with purchase orders from every major hyperscaler, AI cloud and system OEM. NVIDIA’s aggregate supply and capacity commitments increased to $279 billion as of July 26, 2026. This does not translate mechanically into revenue for Coherent or Lumentum, but it demonstrates the scale and duration of upstream commitments being made to support future AI systems. ([investor.nvidia.com](https://investor.nvidia.com/files/content_files/TRANSCRIPT_-NVIDIA-Corp-NVDA-US-Q2-2027-Earnings-Call-26-August-2026-5_00-PM-ET.pdf?utm_source=openai))
The Competitive Landscape Is Broader Than a Two-Company Comparison
Coherent and Lumentum compete with Innolight, Eoptolink, Broadcom, Mitsubishi Electric, Sumitomo Electric, Furukawa Electric, Accelink, Hisense Broadband, Applied Optoelectronics, Source Photonics and other optical suppliers. They also face Cisco’s Acacia, Marvell and Ciena in coherent and DCI technologies; Broadcom and Marvell in DSPs, switching and integrated optical platforms; and numerous Chinese suppliers in transceiver modules. Some competitors are simultaneously customers. A module vendor may purchase Lumentum lasers, Coherent components, Broadcom DSPs and contract manufacturing from Fabrinet, while competing with Coherent or Lumentum in the finished module.
LightCounting data for 2025 estimated approximately $24.1 billion of revenue among major optical-component suppliers, with Innolight at 22.1%, Coherent at 17.2%, Eoptolink at 14.1%, Fabrinet’s optical business at 12.1%, Lumentum at 8.7%, Accelink at 7% and Hisense at 5.5%. These figures should be treated as supplier-revenue shares rather than pure end-product market shares because they mix component, module and outsourced-manufacturing revenue. Nevertheless, the data demonstrate 2 points. Coherent is already one of the world’s largest optical suppliers, while Lumentum’s economic importance is greater than its aggregate revenue share because it occupies scarce upstream component categories. The data also show that Innolight and Eoptolink together represent more than 36% of supplier revenue, giving Chinese vendors greater scale in high-speed Ethernet transceivers than either Western company individually. ([cdn.prod.nntech.io](https://cdn.prod.nntech.io/company-events/reports/c8e95c5e-b786-3dcc-bedb-b29a9e15755c/presentation.pdf?utm_source=openai))
Market share varies considerably by product. TrendForce estimates that Lumentum, Broadcom and Mitsubishi Electric collectively control approximately 72% of the EML market, reflecting the difficult epitaxy, fabrication, modulation and packaging requirements. Lumentum’s relevance is therefore more pronounced in EMLs than in aggregate optical-module rankings. Coherent is one of the few companies capable of challenging that group while also consuming a large portion of its laser output internally. ([trendforce.com](https://www.trendforce.com/presscenter/news/20260603-13077.html?utm_source=openai))
In complete Ethernet transceivers, the Chinese leaders remain formidable. Innolight and Eoptolink operate at scale, focus heavily on the fastest-growing high-speed modules and benefit from dense Asian supply chains. LightCounting reported that Innolight generated $3.3 billion of revenue in calendar Q2 2026 and Eoptolink generated $1.8 billion, both at triple-digit year-over-year growth rates. Coherent has greater materials and component verticality but must compete with those companies on cost, yield, delivery and qualification speed. Lumentum’s Cloud Light franchise gives it participation in modules, but its more defensible position remains the lasers and signal-integrity expertise embedded inside the module. ([lightcounting.com](https://www.lightcounting.com/newsletter/en/september-2026-quarterly-market-update-371?utm_source=openai))
Broadcom is both an ecosystem enabler and a strategic threat. Its optical DSPs, switch silicon, silicon-photonics technology and CPO platforms allow it to influence system architecture. Broadcom’s third-generation 102.4 Tbps Tomahawk 6 CPO switch is designed to reduce optical-interconnect power consumption by approximately 70% relative to traditional pluggables. Broadcom could expand its capture of optical value by combining switching, DSP, SerDes and photonics more tightly, potentially compressing the merchant opportunity for stand-alone module vendors. Conversely, Broadcom’s adoption of CPO validates demand for the high-power external lasers, fibers and optical components that Coherent and Lumentum can provide. ([investors.broadcom.com](https://investors.broadcom.com/news-releases/news-release-details/broadcom-announces-tomahawkr-6-davisson-industrys-first-1024?utm_source=openai))
Marvell is similarly important. Its fiscal Q2 2027 data-center revenue increased 46% year over year, supported by 800G and 1.6T optics, DCI, switching and custom AI silicon. Marvell’s acquisition of Celestial AI added Photonic Fabric technology for optical scale-up connectivity, while its DSP and coherent-module positions extend across scale-out and scale-across networks. The company can compete with Coherent and Lumentum in optical modules and integrated architectures, but it can also purchase their lasers and components. The optical market is therefore unlikely to settle into simple vertical silos. The winning platforms will combine internally developed technology with externally sourced components based on performance, yield and time to market. ([investor.marvell.com](https://investor.marvell.com/news-events/press-releases/detail/1031/marvell-technology-inc-reports-second-quarter-of-fiscal-year-2027-financial-results?utm_source=openai))
Coherent’s Advantage Is Vertical Integration That Now Has a Clearer Strategic Purpose
Coherent’s primary competitive advantage is not merely size. It is the combination of materials science, compound-semiconductor fabrication, passive optics, thermal management, photonic integration and module-level manufacturing. Many optical suppliers are strong at one layer. Coherent can participate from substrate-adjacent processes through a finished transceiver or laser system. This breadth gives customers a single counterparty for increasingly integrated designs and allows Coherent to optimize across interfaces that are otherwise managed by separate vendors.
The move to 6-inch InP manufacturing is the clearest example. Coherent currently operates 6-inch production in Texas and Sweden and plans to begin production in Zurich during the first half of calendar 2027. Management says its 6-inch lines are producing EMLs, CW lasers and photodiodes at yields exceeding those of its 3-inch lines. A 6-inch wafer provides roughly 4 times the device output of a 3-inch wafer at approximately half the wafer cost, before considering the yield advantage. The theoretical cost improvement is unusually large, although the realized benefit will depend on utilization, device mix, depreciation, packaging yields and the pace at which older capacity can be displaced. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
Coherent produced approximately 80% more InP lasers in the June 2026 quarter than a year earlier and expects to double internal InP output capacity year over year one quarter ahead of its original plan. It then plans to more than double capacity again by the end of calendar 2027, with further expansion under consideration. The company’s current bottleneck is InP fabrication rather than transceiver assembly and test. As that bottleneck eases, Coherent should be able to support materially higher transceiver volumes without duplicating all downstream infrastructure. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
Coherent also has manufacturing redundancy that smaller competitors lack. It operates more than 20 production facilities in the United States while retaining a global manufacturing footprint. This is increasingly relevant as customers seek geographic diversification and U.S. policymakers scrutinize Chinese networking equipment and optical modules. The domestic footprint does not make Coherent supply-chain independent, but it makes the company a plausible beneficiary of customer and policy efforts to establish non-Chinese optical capacity. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
The weakness of the model is that vertical integration only creates value if management directs capital toward differentiated layers. A sprawling portfolio can conceal low-return products, duplicate facilities and slow decision-making. Legacy Coherent carried industrial businesses with weaker strategic fit and a cost structure shaped by the II-VI acquisition. The divestitures of aerospace and defense in September 2025 and the Munich materials-processing business in January 2026 indicate that management is willing to shrink the portfolio where ownership does not create sufficient returns. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm?utm_source=openai))
Lumentum’s Advantage Is Precision at the Bottleneck
Lumentum’s competitive advantage is concentrated in the optical source, where small variations in wavelength stability, power, efficiency and signal integrity can materially affect customer yields. Management argues that customers receive better transceiver yields from Lumentum CW lasers because output remains within a narrow performance range. That consistency supports premium pricing even in a product category that might otherwise appear more commoditized than EMLs. The value proposition is not the laser’s percentage of the bill of materials; it is the avoidance of scrapping or reworking a far more expensive module. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Lumentum also appears to have a strong signal-integrity engineering position in 1.6T cloud transceivers. Management believes it is reaching the market ahead of larger competitors in several designs and expects the transition from 800G to 1.6T to intensify through fiscal 2027. This is not yet proof of sustainable share leadership. Optical-module generations are fast, hyperscaler qualifications are concentrated and competitors can close technology gaps. However, the combination of Cloud Light’s module capabilities and Lumentum’s internal lasers gives the company more control over the most challenging electrical-to-optical interfaces than a module assembler relying entirely on external components. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
OCS is Lumentum’s most differentiated systems opportunity. Optical circuit switches route optical signals without repeated optical-electrical-optical conversion, reducing power consumption and allowing network topology to be reconfigured around changing workloads. Lumentum doubled OCS shipments from fiscal Q3 to Q4 2026 and expects its first quarterly OCS revenue above $100 million in fiscal Q1 2027. Management continues to target more than $400 million in the second half of calendar 2026 and expects to become the largest supplier to a major customer during early 2027, despite that customer maintaining an internal source. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Lumentum describes itself as the only merchant supplier currently shipping OCS at this scale. The qualification should be taken seriously but not uncritically. Large hyperscalers have internal switching programs, and multiple optical companies are developing alternatives. Lumentum’s advantage is that it has moved beyond demonstration units into production, software integration and field experience. In a complex system product, installed-base learning can matter as much as component performance. Its planned expansion into higher-port-count, lower-port-count and in-tray OCS systems could extend this lead, although the in-tray opportunity is not expected to become meaningful until approximately 2028. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
AI Networking Is Expanding the Optical Market Across Three Distinct Planes
The first growth plane is scale-out, which connects servers, switches and accelerator clusters across the data center. This remains dominated by pluggable transceivers and is moving from 800G to 1.6T. Despite attention on CPO and NPO, fully retimed pluggables continue to set records. LightCounting expects 800G shipments to more than double in 2026 and 1.6T shipments to reach tens of millions of ports from a small 2025 base. Ethernet transceivers now account for approximately 80% of the total transceiver market. The key implication is that investors should not model CPO as an immediate replacement cycle. Pluggables and integrated optics are likely to coexist for years across different distances and architectures. ([lightcounting.com](https://www.lightcounting.com/newsletter/en/september-2026-quarterly-market-update-371?utm_source=openai))
The second growth plane is scale-across, which connects separate data-center buildings or campuses. AI workloads increasingly require compute, memory and storage resources to be coordinated across physically distributed facilities. This supports coherent DCI, ZR and ZR+ transceivers, optical line systems and high-performance transmission components. Coherent has a broad coherent and multi-rail portfolio, while Lumentum supplies coherent components, tunable modules, amplification and transport subsystems. Marvell, Ciena, Cisco-Acacia and Nokia-Infinera are significant competitors and partners in this layer.
The third and potentially largest incremental plane is scale-up, where hundreds or thousands of accelerators must behave as one tightly connected computing domain. Electrical interconnects become increasingly constrained by reach, bandwidth density and power consumption. NPO and CPO bring optics closer to the switch or accelerator package, reducing electrical trace length and allowing bandwidth to scale without an equivalent increase in power. This architecture creates demand for high-power lasers, external laser modules, optical engines, polarization-maintaining fiber, isolators, passive optics and advanced packaging. It is therefore a content-expansion opportunity for both companies, even if it reduces demand for some traditional pluggable modules.
Coherent and Lumentum agree on the broad timing. Initial CPO-related laser revenue is beginning in scale-out applications, while meaningful scale-up deployments should start later. Coherent expects scale-up CPO revenue to begin in the second half of calendar 2027. Lumentum expects ultra-high-power laser demand for scale-up to ramp in the second half of 2027 ahead of customer deployments in 2028. The apparent one-year lag between component shipments and system deployments is reasonable because customers must qualify, inventory and integrate the optical components before commercial installation. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
The Product Road Maps Are Becoming More Important Than the Current Revenue Mix
Coherent’s near-term growth remains anchored by 800G and 1.6T transceivers. The company expects both generations to grow during calendar 2026, with 1.6T adoption broadening through 2027. It is also expanding OCS, coherent DCI, optical line systems and component sales. Orders extend into calendar 2028, while long-term agreements extend through the end of the decade. This visibility supports capacity investment, although it does not eliminate cancellation, repricing or architecture risk. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
The more important Coherent development is PhotonLink, launched on September 21, 2026. PhotonLink is an integrated-optics platform covering light generation, beam shaping, transmission, detection and electrical conversion. It combines lasers, external laser modules, isolators, polarization-maintaining fiber, attached fiber assemblies, passive components and silicon-photonics PICs. The commercial logic is sound: customers developing CPO or NPO do not necessarily want to qualify and integrate each photonic ingredient independently. A qualified platform can reduce integration time while allowing Coherent to capture more content. The risk is that hyperscalers and switch-silicon vendors may prefer to control the architecture themselves and source components separately. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
Coherent’s Multi-Rail platform is another potential system-level growth driver. It is designed to increase fiber capacity and simplify high-bandwidth optical transport. Management has delivered samples to multiple customers and expects the platform to become a meaningful contributor to revenue and margin over time. Unlike a standard module upgrade, Multi-Rail could move Coherent further into network architecture, increasing content per deployment but also increasing competition with established transport-equipment suppliers.
The company’s industrial technology may also become more relevant to AI. Coherent’s Thermadite material is intended to improve XPU thermal performance, potentially allowing accelerators to operate at higher sustained performance and generate more AI tokens per device. Samples have been delivered to several strategic customers, with revenue expected to ramp in the second half of calendar 2027. This is a credible adjacency because Coherent already possesses the materials and thermal-management capabilities, although customer qualification and system-level integration remain unfinished. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
Lumentum’s immediate roadmap is more concentrated and easier to measure. Its 1.6T transceivers entered production during fiscal Q4 2026, with adoption expected to accelerate through 2027. OCS revenue should exceed $100 million in fiscal Q1 2027. Ultra-high-power laser revenue is expected to reach approximately $50 million in a quarter by the end of calendar 2026 and exceed $100 million in a quarter during fiscal Q3 2027. These products are already shipping, which reduces technology risk relative to longer-dated CPO narratives. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Lumentum has also received its first order for a complete external laser source module. The ELS module packages multiple lasers and supporting optics into a higher-value subsystem for customers that do not want to manage individual laser chips. Shipments are expected around the middle or second half of calendar 2027. The module carries a higher selling price and above-corporate-average margins, although lower margins than stand-alone laser chips. This is strategically significant because it expands Lumentum’s content without requiring it to own the complete CPO optical engine. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
At the technology frontier, Lumentum demonstrated 400G-per-lane EML capability at OFC 2026. Commercial relevance is likely several years away, but the development matters because 3.2T and 6.4T architectures will require either major improvements in per-lane performance or a substantial increase in optical channel count. Lumentum’s ability to remain near the leading edge of EML performance is central to preserving its moat as silicon photonics captures more of the 1.6T market. ([trendforce.com](https://www.trendforce.com/presscenter/news/20260603-13077.html?utm_source=openai))
Suppliers Are a Strategic Constraint, Not a Procurement Detail
InP substrate supply has become one of the industry’s most important bottlenecks. Both companies have secured capacity from AXT. Coherent entered a 3-year development and supply agreement for 6-inch InP substrates and provided approximately $22.3 million of prepayment to support expansion at AXT’s Beijing facility. Lumentum entered a 6-year capacity-reservation agreement, with an initial $43.5 million deposit and a second $43.5 million deposit contemplated for calendar 2028. The larger and longer Lumentum commitment is consistent with the intensity of its merchant laser demand and the slower timing of its new U.S. fab. ([sec.gov](https://www.sec.gov/Archives/edgar/data/1051627/000143774926027677/axti20260630_10q.htm?utm_source=openai))
These agreements de-risk capacity but introduce other risks. AXT’s planned 6-inch expansion is still moving from development and pilot production toward mass production. Yield, crystal-growth scale-up, equipment installation and export permits can affect supply. The location of the manufacturing expansion in China is also strategically awkward at a time of tighter technology controls. Coherent’s internal capacity is in the United States and Europe, but its substrate availability is not fully independent of China. Lumentum has an additional Japanese substrate relationship, which provides some diversification, but management has already indicated that continued demand acceleration could require further supply commitments. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Lumentum’s contract-manufacturing network adds geographic diversification but creates dependency on third-party procurement and inventory management. Coherent’s greater internal manufacturing control reduces some of this exposure, although it still uses contract manufacturers and relies on sole-source or limited-source exotic materials, crystals and optics in parts of the Industrial segment. Vertical integration therefore changes the location of supply-chain risk; it does not eliminate it. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000082031826000020/iivi-20260630.htm))
Management Execution Has Improved at Both Companies
Jim Anderson became Coherent’s CEO on June 3, 2024 after leading Lattice Semiconductor and previously holding senior roles at AMD, Broadcom and Intel. His mandate was to simplify a complex company, improve operating discipline and direct investment toward the highest-growth photonics markets. He refreshed the senior leadership team, appointed Sherri Luther as CFO, reorganized reporting around Datacenter & Communications and Industrial, and divested businesses with weaker strategic fit. ([coherent.com](https://www.coherent.com/news/press-releases/coherent-appoints-jim-anderson-as-chief-executive-officer?utm_source=openai))
The operating evidence since the transition is favorable. Coherent’s pro forma fiscal 2026 revenue increased 28%, Datacenter & Communications grew 40%, the company’s 6-inch InP capacity ramp moved one quarter ahead of plan and customer commitments extended into 2028 and beyond. Management has also been willing to discuss bottlenecks explicitly, identifying InP rather than assembly as the binding constraint. The industrial portfolio remains mixed, but semiconductor and display capital equipment are improving while weaker general-industrial demand has been offset by divestitures and emerging thermal opportunities. ([stockanalysis.com](https://stockanalysis.com/stocks/cohr/transcripts/661756-q4-2026/))
Anderson inherited much of the technology and the early AI demand, so not all recent improvement should be attributed to the management change. Moreover, the next phase is harder than the first. Coherent must translate capacity additions into yield-adjusted output, protect transceiver pricing, integrate PhotonLink into customer architectures and demonstrate that its broad portfolio creates more value than a collection of separate businesses. Capital expenditure is rising materially, and the company will need to avoid building capacity against peak rather than normalized demand.
Lumentum appointed Michael Hurlston CEO effective February 7, 2025, succeeding long-time CEO Alan Lowe. Hurlston previously led Synaptics and Finisar and spent 17 years at Broadcom and its predecessor. Lowe deserves credit for assembling Lumentum’s technology portfolio through Oclaro, NeoPhotonics, Cloud Light and other transactions, but the company entered the AI upcycle after a difficult period of telecom inventory correction, underutilized capacity and weak margins. ([investor.lumentum.com](https://investor.lumentum.com/financial-news-releases/news-details/2025/Lumentum-Announces-Leadership-Transition/default.aspx?utm_source=openai))
Hurlston’s early execution has been strong. Lumentum repriced selected customer agreements, accelerated capacity investment, acquired the Greensboro fab for approximately $38 million, secured substrate capacity, expanded OCS production and converted demand into substantial operating leverage. Non-GAAP gross margin surpassed 50% earlier and at a lower revenue level than management had previously expected. The Greensboro facility gives Lumentum a path to U.S. InP production, although the conversion from GaAs to InP is expected to produce first revenue only in early 2028, with a fuller ramp through 2028 and 2029. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
Lumentum also reduced convertible debt through equitization during fiscal Q4 2026, lowering outstanding convertible debt by approximately $1.1 billion but creating a large non-cash GAAP loss and additional equity dilution. The transaction improved balance-sheet flexibility during a period of aggressive capacity expansion, but it reinforces the need to distinguish operational earnings power from per-share value creation. NVIDIA’s $2 billion investment was structured as convertible preferred stock, whereas Coherent issued common equity. The strategic funding de-risks investment for both companies but permanently changes ownership economics.
New Entrants Can Disrupt the Integration Layer Without Eliminating the Laser Opportunity
Lightmatter is the most credible independent entrant in high-density optical interconnect. It has raised substantial private capital, established partnerships with GUC, Qualcomm, Cadence and Synopsys, and developed Passage optical engines for CPO, NPO and on-board optics. Its L20 engine supports 6.4 Tbps in each direction, while its demonstrated CPO chiplet reached 1.6 Tbps per fiber using a 16-wavelength architecture. Lightmatter is supplying evaluation systems and building manufacturing partnerships intended to support hyperscale deployment. It is no longer a university-style experiment, although broad volume production and field reliability remain unproven. ([lightmatter.co](https://lightmatter.co/press-release/lightmatter-achieves-record-1-6-tbps-per-fiber-to-accelerate-ai-optical-interconnect/?utm_source=openai))
Lightmatter is more threatening to traditional module and optical-engine architectures than to high-performance external laser suppliers. Its platforms still require reliable optical sources, fiber attachment and supporting photonic components. Coherent’s PhotonLink strategy places it closer to direct competition by offering a broad integrated platform. Lumentum can remain a component partner by supplying lasers and ELS modules. The distinction illustrates a broader strategic point: the migration toward integrated optics could compress stand-alone module value while increasing the importance of qualified external light sources.
Celestial AI represents another credible disruptive architecture, although it is now part of Marvell rather than an independent entrant. Its Photonic Fabric is designed to provide high-bandwidth, low-latency optical connectivity at package, system and rack level. Marvell can combine that technology with custom XPUs, switching, DSPs and data-center silicon, creating a more complete platform than a stand-alone photonics startup could offer. This increases competitive pressure on Coherent’s integrated systems and Lumentum’s future ELS modules, but it should also expand demand for the laser sources required to power optical I/O. ([marvell.com](https://www.marvell.com/company/newsroom/marvell-completes-acquisition-of-celestial-ai.html?utm_source=openai))
The Principal Opportunities
The largest opportunity is that optical content per accelerator can rise faster than accelerator shipments. Faster transceivers, additional network tiers, geographically distributed data centers and optical scale-up links all add connections. A transition from 800G to 1.6T does not merely replace one module with another; it can coincide with more ports, more switches, more DCI and new in-rack optical links. This creates multiple concurrent growth curves rather than a single replacement cycle.
Coherent is positioned to capture this through complete transceivers, internal lasers, coherent DCI, OCS, CPO and NPO components, silicon photonics, fiber, passive optics and thermal materials. The company estimates that its existing and emerging product categories can address more than $50 billion of serviceable market by calendar 2030, including over $20 billion of incremental opportunity from newer growth engines. Such estimates are inherently uncertain, but the breadth is strategically credible because Coherent is already qualified at several layers. ([sec.gov](https://www.sec.gov/Archives/edgar/data/820318/000119312526346860/d128030dex992.htm?utm_source=openai))
Lumentum’s opportunity is to remain the scarce upstream supplier even as finished-module architectures change. If 1.6T silicon-photonics modules use more CW lasers, Lumentum can redirect fab output from EMLs toward CW products. If 3.2T returns to heavier EML usage, it retains that capability. If CPO and NPO scale, it can sell ultra-high-power laser chips and complete ELS modules. If hyperscalers expand reconfigurable optical networks, OCS becomes a large systems franchise. This flexibility within the laser layer is more important than defending one specific device architecture.
Geopolitical diversification is an additional opportunity. Chinese suppliers have substantial transceiver share, but Western hyperscalers and governments increasingly view optical connectivity as strategic infrastructure. Coherent has the strongest existing U.S. manufacturing footprint, while Lumentum is building a new domestic InP capability. Any customer-led shift toward dual sourcing or non-Chinese capacity could support structurally higher utilization and stronger negotiating positions, although neither company can immediately replace the scale of the largest Chinese module suppliers.
The Principal Threats
The first threat is cyclicality disguised as structural growth. AI infrastructure is clearly expanding, but optical demand can still overshoot. Long-term agreements and capacity reservations may bring forward orders, while shortages encourage customers to double order. If capacity from Coherent, Lumentum, Chinese laser suppliers and substrate vendors arrives simultaneously, the industry could move from scarcity to excess. Pluggable-module pricing has historically declined with each generation, so volume growth must be judged against annual price erosion and rising capital requirements.
The second threat is customer power. Two customers account for 32% of Coherent revenue and 41.6% of Lumentum revenue. The customers are economically larger, operate internal engineering teams and deliberately cultivate alternative suppliers. Strategic investments from NVIDIA improve alignment, but they also demonstrate that the customer is willing to fund capacity directly in exchange for access rights. Suppliers may earn attractive returns during scarcity while surrendering part of the long-term economics through pricing agreements, capacity commitments or equity dilution.
The third threat is vertical integration by customers and semiconductor vendors. NVIDIA, Broadcom, Marvell and hyperscale cloud companies increasingly co-design switches, accelerators, DSPs, optical engines and network topology. A merchant supplier can lose content if the customer internalizes OCS, integrates modulation into silicon or specifies a modular architecture that turns differentiated subsystems into interchangeable components. Coherent’s response is to broaden its platform. Lumentum’s response is to dominate the hardest component. Both strategies can work, but neither is immune.
The fourth threat is Chinese competition in InP lasers. Lumentum has not yet observed a material impact and argues that emerging suppliers lack demonstrated output, consistency and customer yields. That may be correct today. It is less certain over a 5-year horizon. Chinese module vendors have strong incentives to develop domestic lasers, and government-supported capacity can remain in the market despite initially weak economics. Premium suppliers must continue advancing power, efficiency, lane speed and reliability faster than competitors can commoditize prior generations. ([stockanalysis.com](https://stockanalysis.com/stocks/lite/transcripts/660925-q4-2026/))
The fifth threat is execution. Coherent is simultaneously ramping multiple 6-inch fabs, 1.6T transceivers, OCS, CPO, PhotonLink, Multi-Rail and thermal products. Lumentum is expanding 2 Japanese fabs, converting Greensboro to InP, scaling cloud modules, outsourcing portions of OCS and securing substrate supply. A single technology can perform well in a laboratory and still fail to meet volume yield, reliability, cost or delivery requirements. Current margins assume increasingly efficient production, making yield shortfalls more economically significant.
Comparative Investment Synthesis
Lumentum currently has the stronger operating momentum. It is converting optical shortages into pricing, mix and utilization benefits at an exceptional rate. Its fiscal Q4 non-GAAP gross margin above 50% and operating margin of 36.6% demonstrate scarcity economics rather than ordinary module manufacturing. The near-term milestones are tangible: 1.6T is shipping, OCS is entering quarterly revenue above $100 million and ultra-high-power lasers are moving toward a similar scale. Lumentum is the more direct way to capture the next 12 to 24 months of laser and OCS scarcity.
Coherent has the stronger strategic architecture for a longer holding period. It is larger, less customer-concentrated and positioned across lasers, modules, CPO, NPO, DCI, optical switching, passive components, fiber, silicon photonics and thermal management. Its 6-inch InP manufacturing platform can create a significant cost advantage if the ramp is executed as planned. PhotonLink gives the company a coherent answer to the industry’s move toward integrated optics, while the Industrial segment provides technology adjacencies and some demand diversification.
The trade-off is that Coherent must prove that breadth can translate into better returns rather than permanent complexity. Its margins remain below Lumentum’s, and complete transceiver exposure subjects it to competition from Innolight, Eoptolink and other scaled module vendors. Lumentum must prove that scarcity rents can survive greater capacity, architectural changes and customer bargaining. Its higher concentration, smaller manufacturing footprint and greater dependence on a few bottleneck products make its results more sensitive to qualification losses or design changes.
On balance, Coherent appears better positioned across multiple plausible optical architectures. It does not require EML, CW silicon photonics, pluggables, NPO or CPO to emerge as the single winner because it can supply key content across all of them. Lumentum can generate greater incremental profit if its EML, high-power laser and OCS leadership persists, but a larger proportion of its outcome depends on maintaining premium positions in a narrower set of products. Coherent offers broader participation; Lumentum offers higher sensitivity.
The Scorecard
Coherent has the stronger overall competitive position. Its materials-to-systems integration, 6-inch InP ramp, protocol-agnostic transceiver portfolio, U.S. manufacturing footprint, coherent DCI products, CPO and NPO capabilities, PhotonLink platform and industrial-materials expertise create several independent paths to long-term growth. Management has improved focus through divestitures and has moved critical capacity milestones ahead of schedule. The principal negatives are execution complexity, lower margins than Lumentum, elevated capital intensity and exposure to scaled Chinese transceiver competitors. The long-term thesis depends on vertical integration producing customer value and cost advantages, not merely additional revenue.
Lumentum has the stronger near-term earnings-quality trajectory and the more concentrated exposure to optical scarcity. Its EML and high-power laser capabilities, Cloud Light transceiver platform and merchant OCS lead are difficult to replicate, while management execution since February 2025 has been notably strong. The principal negatives are 41.6% top-2 customer concentration, delayed U.S. InP capacity, dependence on constrained substrates, dilution associated with strategic and debt transactions, and the possibility that silicon photonics or customer internalization shifts economics away from its highest-margin products. Lumentum is the higher-torque beneficiary of the current cycle; Coherent is the more robust owner of the optical stack across a wider range of future architectures.