Meet the Under-the-Radar AI Stock Poised to Outgrow Nvidia

Lumentum is positioned to benefit from AI data-center networking shifting from copper to optical connectivity, with management describing potential demand multiples of roughly 10x for scale-out and 3x-4x above that for initial scale-up deployments beginning in H2 2027. In a fully optical scale-up scenario, management suggested the addressable market could reach 100x the traditional data-center-interconnect market before additional scale-in opportunities. Nvidia's $2 billion March investment in Lumentum reinforces the strategic importance of co-packaged optics and optical components in next-generation AI infrastructure, though the growth case depends on execution and supply capacity.

Analysis

The investable issue is not whether optics penetrates AI fabrics, but where value accrues as architectures change. LITE has exposure to higher-value lasers and indium-phosphide components, but module makers (COHR, AAOI, FN) and switch/DSP suppliers (AVGO, MRVL) can capture a larger portion of the near-term bill of materials before co-packaged optics (CPO) is commercially proven. The likely 6-18 month bottleneck is qualified laser capacity and yield, creating upside for suppliers with validated hyperscaler design wins rather than for the broad optical basket.

The company’s addressable-market framing should not be capitalized at face value: conversion from copper depends on rack topology, power budgets, serviceability, and customer willingness to redesign networks. CPO may reduce pluggable-optics content for some vendors even as it raises laser content, making the critical question LITE’s attach rate and gross-margin capture per switch rather than aggregate optical bandwidth. Near term, AI-networking orders remain vulnerable to digestion if GPU cluster deployments slip; this would hit high-beta component inventories before it meaningfully alters NVDA accelerator demand.

Consensus may be underestimating the timing gap: scale-out optics is a current revenue opportunity, whereas scale-up and scale-in optical adoption require ecosystem standards and field reliability that can extend well beyond the market’s implied schedule. That supports a selective, design-win-driven approach rather than chasing a retail-driven LITE rerating. A decisive catalyst is the next two earnings cycles: backlog conversion, data-center revenue mix, gross-margin progression, and customer concentration disclosures will distinguish real content gains from narrative TAM.

Key Decisions for Investors

• Watch, rather than initiate, LITE until the next earnings report confirms sequential data-center revenue acceleration and stable/improving gross margin; a revenue beat without margin expansion would indicate capacity or pricing leakage and weaken the thesis.
• For a 6-12 month AI-networking allocation, prefer a basket long LITE/COHR versus short an equal-dollar broad semiconductor ETF (SOXX) only after verified optical order guidance; this isolates optical-content upside from a general AI multiple contraction.
• Maintain NVDA as the cleaner 1-3 month expression of AI capex, but do not assume its strategic supplier relationships guarantee LITE volume or economics; monitor disclosed customer concentration and qualification status before treating LITE as a derivative NVDA long.
• Set a risk trigger on LITE and COHR if hyperscaler capex guidance is cut or if inventory days rise for two consecutive quarters; those signals would likely precede a sharp de-rating in component suppliers, whose operating leverage is materially higher than system vendors.
• Avoid long-dated calls on LITE absent visibility into CPO commercialization milestones. The missing inputs are product qualification timing, laser yield, and per-system content; without them, option premium likely prices a structural upside that may not arrive within the contract period.