Lumentum (LITE.US), "standing in the light," is about to announce its performance! Will the short selling of storage and the "" trading add more fuel to the fire?

date
15:53 10/08/2026
avatar
GMT Eight
As the number of GPUs, the scale of model parallelism, and the span of data centers continue to expand, network bandwidth, lasers, optical modules, and CPOs are becoming new marginal bottlenecks.
Lumentum (LITE.US), the core participant in the Alphabet Inc. Class C TPU AI computing power chain and an indispensable supplier of core optical/optical module components in the NVIDIA Corporation and AMD AI GPU-driven AI computing power chain, will announce its Q4 FY2026 earnings after the U.S. market closes on August 11. If Lumentum, which secured strategic investment from NVIDIA Corporation and entered into a long-term optical component procurement agreement, reports stronger-than-expected results and outlook that significantly exceeds market consensus, it would be a major catalyst for the AI super growth stock Lumentum, which has skyrocketed 140% year-to-date, and experienced a meteoric rise of 345% throughout 2025as well as for the Investing in Light theme that has taken over the global stock market recently. The Investing in Light theme refers to the investment boom in the AI data center optical interconnect/optical communication supply chain. According to estimates compiled by analysts on Wall Street, Lumentum's Q4 FY2026 (the second calendar quarter) revenues are expected to be between $960 million and $1.01 billion. The expected NON-GAAP earnings per share are projected to be between $2.85 and $3.05. Consensus estimates from Zacks Investment Research show that analysts expect revenues of about $989 million, which would represent a significant increase of approximately 105.65% compared to the same period last year. The consensus for earnings per share is approximately $2.99, which has been revised upwards by analysts by 3 cents over the past 30 days, in contrast to Lumentum's earnings per share of only $0.88 for the same period last year. As shown in the chart aboveLumentum's stock price has outperformed expectations for earnings per share over the past four quarters, exceeding Zacks' consensus of Wall Street analysts by an average of 9.59%. For a leader in the AI computing power supply chain whose earnings growth trajectory is continually being revised up by analysts, a beat of approximately 10% is already significantly strong. The main reason for Lumentum's extreme stock performance in 2025 lies in the company's ability to benefit from both Alphabet Inc. Class C TPU AI computing power supply chain and NVIDIA Corporation's AI GPU computing power supply chain. Whether it is NVIDIA Corporation's AI GPU infrastructure clusters or Alphabet Inc. Class C TPU clusters (where TPU refers to AI ASIC technology), both rely on the optical interconnect and CPO silicon photonics interconnect capabilities that Lumentum represents. Lumentum has been included in the major U.S. benchmark stock indicesthe S&P 500 and the Nasdaq-100 Index, known as the "global technology stock barometer." In other words, Lumentum has secured passive fund allocation eligibility in both core U.S. benchmark indices, further strengthening its institutional holding base as a core growth stock in AI optical interconnect. The inclusion in benchmark indices has effectively elevated Lumentum from being a niche supplier in AI data center optical interconnect to a core asset in the global technology sector, which also signifies a shift in the main investment narrative for AI computing power from the competition around AI GPU/ASIC single-point computing power to the AI agent-driven full-stack computing power systems. In this narrative shift, the CPU for data centers, storage chips, and the optical interconnect/optical communication supply chain could emerge as the biggest winners. The sustained strength in stock prices of Lumentum and other optical product suppliers for data centers can also be attributed to market rumors that AI chip giant NVIDIA Corporation (NVDA.US) is adjusting its next-generation AI rack computing architecture. This could mean that the next-generation AI computing cluster, Rubin Ultra, may reduce HBM configurations per rack and instead use optical interconnect technologies to link multiple AI racks. With HBM/DRAM memory systems and NAND storage chip prices skyrocketing in recent years, the focus of AI computing infrastructure investment appears to be shifting from GPU/TPU single-point performance and HBM capacity to the overall architecture efficiency of data centers, where high-speed optical interconnect may be the most critical direction. Moving from 800G to 1.6T and 3.2T: Lumentum bets on the optical gain and copper decline trend in AI data centers, entering a supply-demand cycle for high-end lasers. Lumentum's Q4 FY2026 performance is expected to continue benefiting from the strong high-speed optical component demand driven by cloud computing and AI product mix, with both component and system businesses performing robustly. Of the projected 21.3% quarter-over-quarter revenue growth in the current performance guidance, over half is expected to come from optical components for data centers, while the remaining growth is mainly driven by contributions from new high-speed transceivers (i.e., optical module products) and optical switching (OCS). Demand for EML laser chips and scale-across productsincluding pump lasers and narrow linewidth laser componentsmay continue to be particularly strong. In Q3 FY2026, shipments of narrow linewidth lasers increased by over 120% year-on-year, and shipments of pump lasers grew by 80%, with these products effectively already sold out for the foreseeable future of around 2-3 years. Another important catalyst for Q4 FY2026 is likely to come from ramping production of 1.6T transceivers. As Lumentum entered the quarter, management had anticipated increasing shipments of the 1.6T products, including the start of integrating internally sourced continuous wave (CW) lasers; at the same time, continued improvements in yield and reduced scrap rates are expected to help enhance the profitability of the transceiver business. The transition from 800G to 1.6T also brings favorable per-unit value and pricing structure to the company. Lumentum noted that the average selling price (ASP) for the 200G EML used in 1.6T applications is about twice that of the 100G EML currently used for 800G transceivers. Meanwhile, favorable product mix, strict pricing discipline, manufacturing capacity utilization, and operational leverage are expected to further support profit expansion. These factors have driven a 540 basis point quarter-over-quarter increase in non-GAAP gross margins to 47.9% and a 700 basis point increase in operating margins to 32.2% for Q3 FY2026. Despite a favorable demand environment, supply constraints may still limit Lumentum's ability to fulfill customer demand fully in Q4 FY2026. Management indicated that the supply-demand gap for EML exceeds 30%, while supply constraints for pump lasers are even more severe, forcing the company to allocate capacity among different customers. Additionally, as customer demand scales up significantly, the growth of Lumentum's OCS-related product lines, especially those purchased rapidly by cloud computing leaders like Alphabet Inc. Class C, is also constrained by tight supply chain capacity. As illustrated in the chart above, Lumentum (LITE.US) has significantly outperformed its industry peers. Year-to-date, Lumentums stock price has risen by 127.4%, considerably outperforming the S&P IT computer and technology sectors 16.9% increase. The company has demonstrated better performance than optical interconnect device competitors Coherent (COHR.US) and Ciena (CIEN.US), but lagged behind Marvell Technology (MRVL.US). During the same period, the stock price returns for Coherent, Ciena, and Marvell Technology were 81.1%, 72.7%, and 147.6%, respectively. Lumentum and Coherent are competitors in the high-speed optical module and optical transceiver products used for data centers and AI infrastructure optical interconnect. Ciena is a leading provider of optical networking equipment, software, and services among large data centers. Marvell Technology (i.e., Marvell Technology, Inc.) is a competitor in the optical networking space for AI and data center applications, competing with Lumentum, Coherent, and Ciena in AI optical products. Currently, Lumentum's valuation is not cheap. Based on the P/S ratio over the next 12 months, Lumentum currently has a valuation of 10.93x, higher than the overall industry and most comparable peers. The entire industry valuation is approximately 6.54x, with Ciena and Coherent at 7.63x and 6.42x, respectively. However, Lumentum has a significant valuation discount compared to Marvell Technology, whose stock P/S ratio is about 13x. As shown in the chart above, Lumentum (LITE.US) shares are trading at a premium benchmark; the company's growth outlook relies on whether strong AI computing infrastructure demand can persist in the long term. As AI infrastructure increasingly shifts from copper interconnect to optical interconnect systems, the value per optical device is expected to enter years of expansion, placing Lumentum in a favorable position. In addition to its current advantages in the EML and optical transceiver sectors, the company has identified four major growth areasAI data center OCS, optical scale-out for AI data centers, optical scale-up, and optical transceiversmany of which currently contribute relatively limited revenue. Lumentum management anticipates that future contributions from OCS and optical scale-out businesses will become more significant, while optical scale-up could witness significant inflection points starting in 2027. The company has also specifically emphasized the trend of migrating to 1.6T and ultimately 3.2T interconnection technology, which will raise demand for high-speed optical devices in AI training/inference systems. Lumentum boasts a strong market position in differentiated high-performance lasers, while factors such as reliability, performance, and customer manufacturing yields create significant barriers to entry, potentially helping the company capture substantial market share during this expansion phase. Scale-up interconnect and CPO/NPO represent particularly attractive long-term opportunities. Lumentum expects that as optical connections deepen within AI racks and high-speed connections grow between various AI racks, optical scale-up will significantly enhance the intensity of optical device usage and may create market opportunities far larger than scale-out. According to Lumentum management, once scale-up technology is rolled out on a large scale, the optical use intensity could increase by at least threefold; simultaneously, demand for CPO in data centers remains strong, with ongoing growth in demand for CW lasers. As customer collaborations and application scenarios continue to expand, the OCS technology-related product lines also possess immense growth potential, including configurations with higher and lower port counts, as well as potential in-rack applications. In the longer term, Lumentum plans to extend its optical component product portfolio from lasers to include integrated circuits (PICs), photodiodes, laser drivers, and potential transimpedance amplifiers (TIAs), thereby expanding the company's addressable content value in next-generation optical systems. The performance of Applied Optoelectronics has ignited a "buy into AI optical interconnect" trading trend, and Lumentum, which holds positions across the EML to CPO full stack, is likely to further catalyze the optical interconnect space. As Lumentum enters Q4 FY2026, it appears to be backed by strong business momentum, driven by surging optical demand related to AI, accelerated scaling of 1.6T optical transceivers, tight supply for EML and scale-across products, improved product mix, and significant operating leverage. More importantly, many of the company's potential biggest opportunitiesincluding OCS, scale-out CPO, particularly optical scale-upare still in relatively early adoption stages. Lumentums differentiated laser product portfolio, constantly expanding its high-speed optical business in data centers, and deep exposure to the growing demand for AI computing infrastructure interconnections provide a solid foundation for maintaining revenue and profit growth in the long term. The recently disclosed financial results by Applied Optoelectronics (AAOI.US) have provided a very strong demand cross-validation for the entire optical interconnect supply chain. In Q2, AAOI reported revenues of $191.9 million, a year-over-year increase of approximately 86%, with data center revenues rising from $44.8 million in the same period last year to $107.7 million; shipments of 800G products more than doubled compared to the previous quarter, and the company indicated that demand for both 800G and 1.6T is expected to exceed production capacity at least until mid-2027. Notably, it currently has nearly a monthly production capacity of 200,000 units, but plans to expand this to about 650,000 units by the end of 2026indicating that the biggest issue in the current industry is not order volume, but whether enough lasers, DSPs, TIAs, and packaging capacity can be secured to timely manufacture optical modules. The market response following AAOI's earnings report on August 7 was also very straightforward: AAOI rose about 9%, Coherent increased by approximately 13.4%, and Lumentum was up about 6.5%, indicating that funds view AAOI's supply-demand imbalance as a heavy signal that LITE, COHR, and other upstream optical platform companies' high-speed optical device capacities are struggling to keep up with the rapidly rising demand for AI training/inference clusters. Lumentum, which holds multiple positions in EML/CW laser/VCSEL/CPO/OCS, is becoming a very pure optical shovel seller in the AI infrastructure wave. The central reason is that the expansion of AI clusters has shifted from enough computing chips to whether data can move fast enough and efficiently between tens or even hundreds of thousands of GPUs. Lumentum does not merely sell regular optical modules; it positions itself across three levels: on the scale-out side, it offers high-speed EML laser chips as well as 800G/1.6T optical transceivers; on the scale-up side, it advances 1060nm VCSEL arrays, driving optical connections further into AI racks and between racks; on the CPO/silicon photonics side, it provides ultra-high power CW lasers, DWDM external laser sources, and other key light sources; the longer distance scale-across/DCI covers pump lasers, narrow linewidth lasers, tunable lasers, and WSS devices. The 1.6T DR4 module displayed by Lumentum at OFC this year utilizes four 400G differential EMLs and clearly identifies this as a technological stepping stone towards 3.2T; its ultra-high power lasers target the CPO and silicon photonic architecture. In other words, regardless of whether AI data centers are adopting pluggable optical modules, CPO, or ultimately pushing optics into scale-up fabric, Lumentum is positioned to be the most core beneficiary. Recently, market news indicates that NVIDIA Corporation is adjusting its next-generation AI rack computing architecture. The next-generation AI computing cluster, Rubin Ultra, may reduce HBM configurations per rack in favor of using optical interconnect technology to connect multiple AI racks, mainly due to skyrocketing storage chip prices leading to surging storage costs. Meanwhile, leaders in optical devices for AI data center interconnect systems like Lumentum may simultaneously gain port count + single port speed + high-speed optical penetration expansion benefits, which underpin the recent core logic for why optical interconnect themes have outperformed storage chips. As AI models migrate from training to large-scale inference, MoE, and Agentic AI, collective communication, KV/cache migration, and model parallel communications between computing nodes will amplify east-west traffic. Additionally, the transition from 800G to 1.6T and continuing towards 3.2T development faces limits in insertion loss, power consumption, and signal integrity at higher data rates and longer distances, making some of the scale-up connections previously managed by copper cables increasingly viable for optics. NVIDIA Corporation has directly integrated silicon photonics into its Spectrum-X/Quantum-X switches, claiming that a 1.6Tb/s-level optical switching architecture offers higher bandwidth density, lower power consumption, and greater network reliability, while Lumentum is accelerating its VCSEL scale-up solutions directly packaged with host ASICs. The data center storage supply chain has experienced extreme price surges and profitability explosions, and the market seems to be trading on the expectation of slowing future ASP growth; however, AI optical interconnect is progressing from 800G to 1.6T and continuing from scale-out to scale-up/CPO, thus profit forecasts remain in an accelerating upward revision phase. In other words, as the number of GPUs, the scale of model parallelism, and the span of data centers continue to expand, network bandwidth, lasers, optical modules, and CPO are becoming the new marginal bottlenecks. The market does not view storage chips as becoming unimportant; rather, it is trading based on which supply bottleneck is becoming increasingly tight, a logic particularly suited to explain the recent trading trend of shorting storage and buying optics.