Nvidia's Latest AI Partner Was a Bargain Hiding in Plain Sight. Is It Still a Buy?

Corning's optical communications business is accelerating sharply, with segment net income up 28% in 2024, 71% in 2025, and Q1 2026 net sales up 36% year over year while net income jumped 93%. The article argues AI-driven demand for fiber-optic cables is the key growth driver, but warns that manufacturing capacity is currently maxed out despite Nvidia helping fund three new facilities. Corning’s 259% stock gain over the past year reflects this AI tailwind, though the piece flags valuation and supply constraints as risks.

Analysis

The market is likely underestimating how much of the AI buildout shifts value from compute vendors to the interconnect layer. If fiber is the bottleneck, the next marginal dollar of AI capex can increasingly flow to suppliers of cables, connectors, and optical tooling rather than to the hyperscalers alone. That creates a second-order winner set in the supply chain, but it also means the equity re-rating can front-run actual revenue realization by several quarters.

The key risk is that the current narrative assumes demand is infinitely elastic while capacity is not. In practice, when a supplier hits utilization ceilings, customers either defer deployments, substitute product mix, or negotiate harder on price once competing capacity comes online. That suggests GLW’s earnings power may look strongest over the next 1-2 quarters, then normalize if new facilities arrive into a less-urgent order book or if supply-chain bottlenecks migrate upstream.

Consensus appears to be treating GLW as a durable AI monopoly, but the more defensible view is that this is a cyclical scarcity trade wrapped in a structural story. Scarcity premiums can be very large, yet they often compress quickly once the market prices in a 12-24 month supply response. The more interesting asymmetry may be in NVDA, which benefits from enabling the capex cycle but is also effectively underwriting competitor capacity expansion that could dilute its own ecosystem control over time.

The less obvious loser is the subset of AI infrastructure vendors whose margins depend on proprietary bundling rather than pure throughput. If fiber becomes the universal bottleneck, customers gain leverage to standardize around the most scalable interconnect solutions, which can pressure lower-quality copper or niche networking substitutes. That makes this theme less about one winner and more about a repricing of the entire data-center plumbing stack.

Key Decisions for Investors

• Buy GLW only on pullbacks or after a capacity-update catalyst; near term 1-2 quarters are likely strongest, but upside beyond that depends on evidence that new facilities translate into sustained backlog rather than peak-margin supply relief.
• Pair long GLW / short a weaker networking or copper-interconnect proxy over 3-6 months; the trade captures interconnect share gain while reducing exposure to broad AI beta. Stop if GLW backlog growth decelerates or competitors announce faster capacity additions.
• Stay structurally long NVDA on any weakness, but avoid chasing strength after supply-chain partnership headlines; the better entry is on air pockets, since the market may over-discount the idea that NVDA is funding future capacity that could normalize bottlenecks.
• Avoid initiating new longs in the AI infrastructure basket after a 1-year 259% move in GLW; use out-of-the-money call spreads instead of outright stock if you want upside exposure with defined downside.
• If you already own GLW, trim into strength and hedge with short-dated puts around the next earnings/capex disclosure, because the primary risk is not demand collapse but a supply-response surprise that compresses the scarcity premium.