Optical Networking Is AI's Most Overlooked Bottleneck. Here's Why This Nvidia-Backed Company Is the Stock to Own.

Coherent reported fiscal 2025 revenue growth of 23%, and analysts expect revenue to rise at a 31% CAGR through fiscal 2028 with EPS turning positive in fiscal 2026. The company is benefiting from AI-driven demand for 800G, 1.6T and next-generation 3.2T optical networking products, plus $2 billion of Nvidia investment and $50 million in CHIPS Act funding for its Sherman, Texas plant. Shares have already rallied about 30% since Nvidia’s investment, and the article argues the stock still has long-term upside despite trading at 65x next year’s earnings.

Analysis

The important read-through is not just that AI optics demand is rising, but that the bottleneck is moving from compute to interconnect. As GPU clusters scale, the value pool shifts toward firms that can solve bandwidth, power, and thermal constraints; that favors COHR as a levered beneficiary of the next wave of datacenter capex, while modestly improving the optics mix for NVDA’s ecosystem by reducing deployment friction. The second-order winner is likely the supply chain around high-speed packaging, test equipment, and photonics materials, where capacity tightness can persist longer than end-demand headlines.

The market may still be underestimating how quickly optics can re-rate when product cycles move from 800G to 1.6T and then 3.2T. That transition is structurally different from prior cyclical upswings because it is tied to AI cluster power density, not generic bandwidth growth; once deployments standardize around fiber, the attach rate becomes stickier and customer switching costs rise. The risk is that valuation already discounts several years of execution, so any pause in hyperscaler orders or delay in standards migration could compress multiple before fundamentals roll over.

Near term, COHR is a momentum-long but not a clean “buy and forget” name: the stock likely trades on order-book visibility and AI capex sentiment over the next 1-2 quarters, while actual earnings leverage matters more over 12-24 months. The contrarian angle is that consensus may be overconfident that every AI infrastructure dollar accrues to the optics layer; if GPU supply, networking integration, or power constraints force slower cluster buildouts, near-term upside could be more limited than the narrative implies. NVDA benefits indirectly, but the cleaner expression of this theme remains the picks-and-shovels supplier with operating leverage to optical content expansion.

Key Decisions for Investors

• Long COHR on 3-6 month horizon; use pullbacks to build, with upside tied to hyperscaler order confirmation and 1.6T adoption. Risk/reward is attractive only if you believe AI datacenter capex stays >2026 trend; otherwise multiple compression can dominate.
• Pair trade: long COHR / short a slower-growing industrial laser or legacy optics peer over 3-9 months. Thesis: AI-specific optical mix should outgrow cyclical exposure, and the market will keep paying up for visible datacenter content.
• Buy NVDA vs short an AI hardware basket with less direct interconnect exposure for 6-12 months. If the market shifts from compute scarcity to system integration, NVDA remains core, but optics beneficiaries should catch incremental spend.
• Sell upside calls on COHR against a long stock position into strength. Valuation is already elevated; covered calls monetize implied volatility while preserving participation if AI optics rerate further.
• Set a catalyst watch for next hyperscaler capex commentary and Coherent order backlog updates over the next 1-2 quarters; if visibility weakens, trim COHR quickly because the multiple can de-rate faster than estimates revise.