▶ See it in 3D Photonics region
Silicon cannot make light. Every optical link starts with an indium-phosphide laser — fewer than five firms make them at volume, and NVIDIA locked up ~$4B of Lumentum & Coherent supply through 2027.
To send data as light you first need a laser — and lasers can't be made from ordinary silicon, so they come from a tiny club of specialists. Demand is running so far ahead of supply that NVIDIA pre-paid ~$4B just to lock it up. When something is this scarce, the few suppliers get real pricing power.
Every optical link begins with a laser — and silicon cannot make light. EML/CW lasers are grown on indium-phosphide (InP), not silicon, at 15–50% yields. Fewer than five firms make them at volume.
The single tightest link in optics. Lumentum is the only vendor shipping 200G/lane EML lasers at volume (needed for 1.6T), and all EML capacity is under long-term agreements through 2027. NVIDIA put $2B into each of Lumentum and Coherent in March 2026 — capacity capture disguised as investment; McKinsey models a 30–40% shortfall in 1.6T devices through 2029. Everyone else gets rationed — including China, whose domestic 25G+ laser localisation is under 15%.
Physics, not factories: silicon cannot emit light, so lasers must be grown on indium-phosphide wafers at 15–50% yields — capacity can't simply be bought. Fewer than five firms do it at volume, and NVIDIA pre-paying ~$4B to lock supply through 2027 shows exactly how binding that is.
The only vendor shipping 200G/lane EML lasers at volume — the exact part 1.6T optics needs — with all EML capacity under long-term agreements through 2027. NVIDIA invested $2B (Mar 2026) to lock it. Purest laser-chokepoint play.
~25% transceiver share + vertically integrated InP lasers, with all EML capacity locked under long-term agreements through 2027. NVIDIA invested $2B (Mar 2026) explicitly to expand indium-phosphide capacity and pre-allocate it.
InP/CW lasers, fiber, connectors. Stock +85% in 2025 on laser tightness. Japan's component grip.