Lumentum is undershipping optical transceiver demand by approximately 30%, with all electro-absorption modulated laser (EML) capacity committed through 2027. The company's optical component systems (OCS) order backlog has surged past $400 million, with most orders scheduled for second-half delivery.
The supply constraint centers on EML transceivers, which dominate initial 1.6 terabit designs. Adoption of 200-gigabit lane speeds is exceeding industry expectations, driving demand beyond manufacturing capacity. Lumentum increased indium phosphide production capacity by over 20% in the December quarter, with further expansion planned.
The company achieved 65% year-over-year revenue growth while implementing pricing increases that boosted both revenue and gross margins. Long-term supply agreements now lock in capacity through calendar year 2027, leaving minimal room for new customer orders.
Optical transceivers connect servers and networking equipment in datacenters, with higher-speed components required for AI training and inference workloads. The transition from 800-gigabit to 1.6-terabit transceivers coincides with massive AI infrastructure investments by hyperscalers.
Supply constraints extend across the optical component supply chain. Competitors including II-VI and Coherent face similar capacity limitations as datacenter operators place orders years in advance to secure supply. The backlog duration indicates suppliers cannot meet demand without substantial manufacturing expansion.
The bottleneck could limit how quickly companies can deploy AI infrastructure. Microsoft, Google, Amazon, and Meta are racing to build AI datacenter capacity, but transceiver availability may throttle their timelines. Each rack of AI servers requires dozens of high-speed optical connections.
Indium phosphide wafer production represents the core constraint. The semiconductor material enables high-speed optical components but requires specialized fabrication facilities. Building new capacity takes 18-24 months, creating a structural shortage that pricing alone cannot resolve.
Pricing power has shifted decisively to component suppliers. ASPs are rising across the optical transceiver market as customers accept higher costs to secure allocation. Gross margins are expanding as suppliers prioritize higher-speed, higher-margin products for AI applications.
The shortage validates predictions that physical infrastructure—not algorithms or chips—would constrain AI scaling. Datacenter operators who locked in optical component supply early gain competitive advantage in the race to deploy AI capacity.

