Nvidia CEO Jensen Huang Highlighted a New AI Bottleneck. 3 AI Stocks That Could Benefit.

Nvidia is targeting the AI bottleneck of faster data transfer, investing $2B in Coherent, $2B in Lumentum, and $1B in Nokia. The article highlights strong operating momentum at the investees—Coherent revenue +27% to $1.8B (EPS +55%) and Lumentum revenue nearly doubling to $808.4M (components +77%, systems +121%)—but notes concentration risk at Coherent (data center/communications ~75% of revenue). With shares up roughly +110% (Coherent) and +130% (Lumentum) and Nokia cloud/AI revenue +49% in Q1 2026, sentiment is constructive though expectations look high given Lumentum’s ~50x forward P/E.

Analysis

This is less about one-off endorsement and more about Nvidia validating a new AI bottleneck: bandwidth, not compute. That shifts incremental capex toward optics/photonics, which should expand the revenue pool for COHR and LITE, but the first-order market reaction will likely outrun actual fundamentals because qualification cycles in data centers are measured in quarters, not weeks. In the next 1-3 months, the trade is mostly sentiment and order-book optionality; the real earnings impact should show up 2-4 quarters out if hyperscalers keep spending.

The second-order winner is anyone selling picks-and-shovels into packaging, lasers, and network optimization, but the better relative-risk setup may be among the suppliers with more diversified end markets and lower starting valuations. LITE has more room to disappoint given stretched expectations, while COHR has already started to monetize the theme through data-center exposure; NOK is a longer-dated execution story that could rerate only if AI networking becomes a meaningful, recurring revenue line rather than a press-release narrative. A likely loser is the cohort of AI names whose bull case assumes perpetual GPU scarcity without considering that faster interconnect can ease system constraints and shift bargaining power back to platform buyers.

The contrarian view is that Nvidia may be securing supply for itself more than creating a durable public-market windfall for these vendors. If that’s right, margin expansion will be capped by customer concentration, R&D intensity, and pricing pressure once capacity comes online. Falsify the thesis if COHR/LITE backlog and data-center revenue do not inflect by the next two reporting cycles, or if hyperscaler capex guides flatten; in that case the theme becomes a multiple trade, not a fundamentals trade.

Key Decisions for Investors

• Go long COHR on pullbacks over the next 1-3 weeks; target 10-15% relative outperformance vs. SOXX over 3-6 months if data-center orders convert, and cut if next print shows backlog or margin disappointment.
• Pair trade: long COHR / short LITE for 1-3 months to express the more durable balance-sheet and valuation setup; thesis breaks if LITE continues to beat on gross margin and raises guidance again.
• Avoid chasing NOK after the move; treat it as a watchlist name only. Re-enter only if AI/networking revenue becomes a larger share of the mix and the stock de-risks on execution, otherwise the stock remains a story-first, fundamentals-later trade.
• Use NVDA as a hedge, not the main expression: if the market overprices optics beneficiaries, shorting a small basket of the most extended photonics names against a modest NVDA long can isolate the capex-validation effect without taking broad AI beta.