EO/IR & thermal sensors

The choke is two layers deep

Every Western ISR gimbal and interceptor seeker stares through the same bottleneck twice: once at the lens (germanium long-wave IR optics) and once at the detector (the focal-plane array behind it). China refines roughly two-thirds of the world's germanium and banned its export to the US outright in December 2024. The November 2025 trade truce suspended that ban only until November 27, 2026 — and explicitly kept the military end-user prohibition, so defense optics buyers remain effectively embargoed. Spot germanium has gone from ~$1,200/kg in early 2023 to roughly $8,600/kg by late May 2026 on the strategic-metals index — up nearly 50% in 2026 alone, though Western in-warehouse assessments ran closer to $5,800/kg as recently as April. The material escape valves are few: Teck's Trail smelter is North America's only commercial-scale primary germanium refiner, and Umicore's Electro-Optic Materials unit is the key Western optics-grade processor and recycler.

At the detector layer, merchant US supply of defense-grade thermal cores concentrates in two names. Teledyne FLIR's Lepton/Boson/Hadron families are the default core in Western drone payloads; Leonardo DRS runs merchant domestic IR focal-plane-array fabs in Dallas, spanning uncooled VOx bolometers and cooled MCT for the Army's third-gen FLIR. France's Lynred is the other Western volume producer, and RTX, Lockheed, L3Harris and BAE run captive detector lines for their own programs — but for a US drone payload buying cores on the open market, the funnel narrows to TDY and DRS. The moat isn't IP on paper — it's twenty-plus years of wafer-level vacuum-packaging yield learning, ITAR fences, and NDAA/Blue-UAS compliance rules that keep the Chinese volume leaders (InfiRay, Hikmicro) out of US defense programs. Cores get designed into payloads and re-qualification cycles run years, so share doesn't move between contract cycles.

What it takes to break it

Two vectors. Substitution at the lens is live: LightPath's BlackDiamond chalcogenide glass (exclusive Naval Research Laboratory license) has already moved a defense prime off germanium and is shipping in germanium-free cooled G5 cameras, with a potential Lockheed missile-program ramp behind it — contingent on Lockheed's NGSRI bid winning the Army's two-way Stinger-replacement downselect, with production around 2028. Note what that breaks: the material choke, not the device choke — de-risking the input actually entrenches the US core makers. Competing at the detector means a greenfield FPA fab, years up the yield curve, and program qualifications nobody funds on spec. Until someone does, the duopoly prices like infrastructure.

What forces the reprice

Near term, Teledyne's May 2026 LASSO award for Rogue 1 is underappreciated: the core maker just moved downstream into the munition itself, capturing seeker, optics and airframe margin in one SKU. The award is an initial test-and-evaluation buy — up to 130 systems over two years — so the margin story builds over quarters, not in the next print. The structural kicker is that counter-UAS doctrine turns thermal cores into consumables: every expendable interceptor burns a seeker on impact (Rogue 1 itself is marketed as recoverable, but the doctrinal direction is expendable mass). And the binary sits on November 27, 2026 — if the suspension lapses without a deal, germanium snaps back to full embargo, spot rips, and everything with a de-China'd IR supply chain (LPTH, DRS, TDY) reprices in a week. Own the layer the Pentagon cannot requalify away from.