Lumentum Holdings Inc. (LITE.US), a core participant in the "Google TPU AI computing chain" and an indispensable supplier of optical components and transceivers for the "NVIDIA and AMD AI GPU computing chains," plans to soon establish a new, large-scale manufacturing facility in Greensboro, North Carolina. This factory will produce advanced lasers for leading AI computing infrastructure companies such as NVIDIA, AMD, and Cerebras Systems, as well as for the world's largest AI data centers built by Microsoft, Meta, and Google. The 240,000-square-foot facility is already operational and will be adapted to manufacture Lumentum's indium phosphide (InP)-based data center optical interconnect products. These include key optical components for large-scale AI data centers, such as continuous wave lasers and ultra-high-power lasers. Volume production ramp-up is expected to begin by mid-2028. NVIDIA (NVDA.US), the dominant force in AI chips, is set to be the first major customer for this new factory.
Lumentum stated that the site was acquired from another US semiconductor manufacturer, Qorvo (QRVO.US). The location was chosen due to the local availability of a highly skilled workforce, well-developed infrastructure, and supportive economic development environment at both the federal and state levels. The acquisition agreement includes the transfer of an experienced semiconductor manufacturing team, which will enable Lumentum to accelerate the expansion of production capacity for optical interconnect products, which are urgently needed in the global wave of AI data center construction, and to efficiently increase actual output.
Lumentum's CEO, Michael Hurlston, commented, "Our customers are building the AI computing infrastructure that will define the next generation of supercomputing. Adding this InP manufacturing facility will significantly expand our production capacity, deepen our strategic partnerships, and ensure we can deliver the performance, reliability, and scale required for the AI technology revolution." Lumentum plans to invest hundreds of millions of dollars over the coming years to expand the factory's capacity and continue enhancing its advanced manufacturing capabilities, while also preserving and creating over 400 high-end US manufacturing jobs.
Lumentum is positioned to fully capitalize on the AI infrastructure boom. Regardless of whether NVIDIA's GPU or Google's TPU technology paths gain dominance, the company is expected to continue its bull market trajectory. An increasing number of Wall Street institutions are bullish on Lumentum's stock, anticipating a strong continuation of its 2025 bull run. Lumentum's stock has surged an impressive 340% in 2025, solidifying its status as a top-performing AI computing theme stock globally, with a remarkable 110% gain year-to-date.
The primary reason for Lumentum's exceptional stock performance lies in its ability to benefit simultaneously from both the Google TPU AI computing supply chain and the NVIDIA-led AI GPU computing supply chain. Future AI computing infrastructure clusters, whether based on NVIDIA GPUs or Google TPUs (which represent the AI ASIC technology path), will rely heavily on the high-speed optical and silicon photonics interconnect capabilities that Lumentum provides. A core commonality in these AI super-clusters is the need for extremely high bandwidth, ultra-low latency, and high energy efficiency for internal data center interconnects. Traditional copper cables or electronic switch solutions struggle to meet demands when scaling to thousands or even tens of thousands of chips due to explosive increases in power consumption and heat loss.
Optical interconnect technologies, including Co-Packaged Optics (CPO), silicon photonic switches, and Optical Circuit Switching (OCS), use light signals instead of electrical signals, significantly enhancing bandwidth density and energy efficiency in large-scale AI training and inference networks while reducing latency and power consumption. This demand for greater optical interconnect capacity is common to both GPU and TPU clusters. For instance, NVIDIA's Spectrum-X/Quantum-X silicon photonic network switches explicitly integrate laser and photonic technologies to improve power efficiency and network capacity, with Lumentum's high-performance lasers and optical components being essential parts of these switches. Google has already integrated OCS clusters into its Jupiter/AI data center network architecture on a large scale to support its TPU AI systems and massive training/inference systems. Lumentum's R300/R64 OCS products are specifically designed for large-scale cloud and AI/ML data center networks, using MEMS-based optical circuits to establish direct light paths between endpoints, bypassing intermediate electrical switching and OEO conversion, and emphasizing high port count, low latency, and low power consumption.
Looking ahead, particularly within NVIDIA's AI GPU computing clusters, Lumentum is well-positioned to be a primary beneficiary due to its capital investments, production capacity, and R&D partnerships, especially in key optical interconnect areas like UHP lasers, External Laser Sources (ELS), CPO, and OCS. Lumentum is highly likely to become a critical supplier in the upcoming AI optical interconnect wave, with particular importance for the NVIDIA GPU ecosystem.
In a recent investor note, a team led by Stifel analyst Ruben Roy stated that Lumentum presented a long-term quarterly revenue run-rate expectation of up to $2 billion at the OFC conference, making the company's stock outlook more optimistic than ever. This projected expansion level, equivalent to an annualized revenue expectation of approximately $8 billion, represents a much stronger quarterly and annual outlook compared to current performance. For the quarter ending December 27, 2025, the company reported revenue of $665.5 million, a significant 65.5% year-over-year increase, reaching the high end of its guidance.
The overall trend in data center optical interconnect development observed at OFC 2026 is very clear: the bottleneck for hyperscale AI data centers is shifting from single-chip computing power to "cluster-level interconnect bandwidth, energy consumption, and scalability." Lumentum is a key manufacturer of critical optical communication components (lasers, optical device supplier) and also supplies some optical transceiver products and subsystems. However, its core technological advantage lies in its unique ability to build the fundamental light source and high-speed optical device platforms, which are extensively integrated into various optical transceivers and communication systems within large AI data centers.
Lumentum is not a platform player like Broadcom, which focuses on switch ASICs/DSPs, nor is it a broad-spectrum optical company like Coherent that presents a full stack including silicon photonics, VCSELs, and InP-on-silicon. Lumentum's true strength, and what attracts Wall Street investment, lies in its expertise in the most challenging aspects of AI optical interconnect: InP lasers, EMLs, Ultra-High-Power (UHP) lasers, External Laser Sources, and OCS. These areas are the most difficult to scale, achieve high yields in, and are most prone to technological bottlenecks.
The latest product releases from the three optical giants—Broadcom, Coherent, and Lumentum—all strongly point to the same developments: 1.6T technology is moving from R&D to the verge of mass production, 400G per lane is becoming the new physical layer foundation for scaling towards 3.2T, 12.8T, and even larger AI clusters. The most representative technologies of "optical power," namely CPO and OCS, are increasingly transitioning from optional to necessary components for large-scale AI fabric—high-performance network architectures optimized for AI data center workloads.
Lumentum anticipates that its EML production capacity will increase by over 50% by the end of fiscal year 2026 compared to 2025. The company has already advanced its indium phosphide (InP) production expansion plan by approximately 40% and forecasts an 85% compound annual growth rate (CAGR) in AI data center demand for InP through 2030. Lumentum expects the CAGR for OCS shipments to exceed 150% between 2025 and 2028, targeting over $1 billion in annualized revenue by 2027.