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

Lumentum is positioned to benefit from AI data-center networking demand as optical interconnects increasingly replace copper connections for high-speed scale-out, scale-up and eventually scale-in architectures. Management indicated scale-across and scale-out optical opportunities could each be roughly 10x the traditional data-center interconnect market, while scale-up could add another 3x-4x initially and potentially 10x in a fully optical scenario. Nvidia's $2 billion March investment underscores the strategic importance of Lumentum's optical technology, though the article's growth assumptions remain long-dated and execution- and supply-dependent.

Analysis

The investable implication is not simply higher optical spend, but a change in the bottleneck: cluster utilization increasingly depends on bandwidth, latency, and power per bit rather than incremental GPU availability. LITE and COHR have the highest direct exposure to indium-phosphide lasers and photonic components, while FABR can benefit from higher-value module assembly without bearing the same laser-yield risk. The first earnings impact should come from 800G/1.6T interconnect mix over the next 1-3 quarters; the more consequential scale-up optical opportunity is a 2027-28 architecture transition and should not be capitalized at full value today.

The article's addressable-market arithmetic is promotional rather than a revenue forecast: optical content can rise sharply while unit pricing falls, customer concentration increases, and adoption remains limited to the largest AI fabrics. CPO is particularly double-edged for LITE: it expands laser content but can reduce the addressable market for conventional pluggable transceiver suppliers and shift value toward switch/ASIC integration at AVGO and NVDA. Copper-linked connectivity vendors such as CRDO face a relative risk only if optics penetrates very short-reach links faster than expected; near term, active electrical cables can remain the lower-cost bridge.

Consensus is likely too quick to treat strategic validation as a guaranteed volume commitment. The key falsifier is not a technology demo but sustained datacom revenue growth, gross-margin expansion from product mix, and evidence that supply yields can support deliveries without expedited-cost pressure. A hyperscaler capex pause, slower AI-cluster deployment, or a delay in CPO standardization would de-rate optical names quickly because expectations are already tied to a multiyear inflection.

Key Decisions for Investors

• Watch, do not chase LITE on promotional TAM framing; initiate only after the next earnings cycle confirms datacom order growth and gross-margin progression. Use a 12-18 month horizon, with thesis invalidated by two consecutive quarters of weak datacom guidance or margin deterioration despite revenue growth.
• Construct a 6-12 month relative-value basket: long LITE and COHR, funded by a short in CRDO, sized modestly. This expresses rising optical-content intensity while reducing broad AI-capex beta; cover the CRDO short if active-electrical-cable demand remains above guidance or optical deployment timelines slip beyond 2027.
• For a lower single-name-risk expression, accumulate SOX exposure only on broad semiconductor pullbacks rather than treating LITE as a standalone AI proxy. NVDA/AVGO system shipments remain the gating variable for optical volumes, so a deterioration in their networking or AI-system backlog is the earliest warning signal for the optical thesis.
• Set an event alert around hyperscaler capex guidance and LITE/COHR book-to-bill disclosures over the next 1-3 months. Upgrade the optical overweight only if demand visibility extends beyond a single customer or product generation; concentration and qualification risk otherwise warrant a discounted multiple.