Corning Incorporated

Corning is a genuinely exceptional materials science franchise trading at a genuinely stretched multiple. The optical fiber moat is real — the fusion draw process is harder to replicate than most investors appreciate, hyperscaler supply agreements are long-dated with take-or-pay structures, and AI data center demand is architecturally different from the telco overbuild of two decades ago because this fiber is densely co-located rather than running dark underground. But a real moat and a buyable price are separate questions. A capital-intensive manufacturer with a multi-year ROIC history averaging below cost of capital is not a software business, and paying software-adjacent multiples requires a conviction in permanently elevated returns that the track record simply does not support. The trajectory is improving and the improvement is honest — operating leverage driven rather than buyback-engineered. Optical Communications is becoming the dominant segment with expanding margins as high-density enterprise fiber commands better pricing than commodity carrier glass. Springboard targets have been raised and met ahead of schedule, which is the right kind of management surprise. Display will drag and Environmental faces structural EV headwinds, but Optical's momentum can absorb those if the hyperscaler agreements close as indicated. The single biggest risk is co-packaged optics — the architectural shift that embeds photonic interconnects directly into switch silicon, reducing the fiber intensity per unit of compute. If AI data center design migrates toward CPO at scale, Corning's optical tailwind reverses sharply, and the multiple collapses back toward historical norms at a much lower earnings base. Management explicitly excluded CPO revenue from their forecasts — intellectually honest, but it also means the market is pricing in an AI fiber supercycle that could be disrupted from within the same technology ecosystem driving it.