Optical-network customers are reserving compound-semiconductor capacity years ahead
United States, China, United Kingdom, Israel, Japan, and Global
Share ↗
The signal
Large customer prepayments, minimum-volume agreements, and multi-year capacity reservations for indium-phosphide substrates and silicon-photonics production indicate that optical data movement is creating a strategic upstream materials and foundry layer before next-generation network architectures reach full volume.
Why this one
The selection explanation is unavailable for this edition.
The clue
Customers are no longer merely forecasting demand for optical-network components. They are making substantial prepayments and minimum-volume commitments for indium-phosphide substrates and silicon-photonics production.
Why it matters
AI is one cause, but the broader constraint is data movement: processors, memory and distributed computing systems need faster links without proportionally increasing power consumption. This could elevate compound-semiconductor substrates, photonic foundries, lasers and packaging from specialist components into strategic infrastructure dependencies.
Evidence and interpretation
- Observed facts: Lumentum agreed to deposits totaling as much as $87 million to reserve AXT indium-phosphide substrate capacity through 2031. Tower Semiconductor reported $290 million in customer prepayments associated with $1.3 billion of 2027 silicon-photonics contracts. IQE and Tower separately established minimum-volume commitments for indium-phosphide epiwafers supporting 200Gb/s-per-lane production and 400Gb/s-per-lane prototypes.
- Interpretation: Customers are securing the upstream substrate and foundry layers before future optical architectures are fully settled. That is a stronger trend indicator than product-roadmap claims alone.
- Speculation: If these reservations become sustained shipments, optical connectivity could become a limiting infrastructure layer—and currently obscure compound-semiconductor suppliers could gain negotiating power.
What to watch next
The thesis strengthens if deposits convert into rising wafer shipments, 6-inch InP yields improve, and 200G-per-lane or co-packaged optical products enter repeat production; it weakens if customers cancel reservations, optical buildouts slow, or alternative laser and modulator technologies materially reduce InP content per connection.