Infrared Thermal Imaging Market 2026: Germanium to Chalcogenide, 8 µm vs 12 µm

INFRARED & INVISIBLE LIGHT · IR OPTICS & MATERIALS · OCTOBER 2026

The 2026 read on infrared is a market split in two directions at once: defence buyers are chasing megapixel-class thermal imaging for drones, counter-drone systems and border surveillance, while consumer and industrial buyers push the opposite way toward cheap, tiny, low-resolution IR sensors. The decision that lands on your desk is narrower and more practical — 8 µm or 12 µm pixel pitch, and whether the lens in front of it is made of germanium or chalcogenide glass.

MEMS Consulting (麦妆姓咨询) summarised the 2026 edition of Yole Group’s Thermal Imaging and Sensing report on 28 September 2026. The headline is that defence is the engine of infrared growth right now. The accelerating use of unmanned aerial vehicles, counter-UAS systems and border surveillance has pulled the microbolometer supply chain forward and created the sort of unit volumes that decide how a market develops.

Automotive is the opposite case. The report is explicit that the segment has not released its potential: the growth logic still exists, but volume adoption keeps being pushed to the right, so automotive contributes a delayed rather than a cancelled share of growth. Industrial, consumer and medical demand continues on already-validated applications — steady, resilient, but not enough on its own to bend the curve.

8 µm versus 12 µm: a lens decision disguised as a sensor decision

The most useful paragraph in the report for anyone buying infrared optics is about pixel pitch. The industry is diverging. Chinese manufacturers are pushing 8 µm pixel pitch to cut cost and shrink module size, while Western manufacturers are holding the 12 µm standard.

The Western argument is not sentimental. Smaller pixels require a larger-aperture optical system and tighter manufacturing tolerances, and the report notes that this can end up raising total system cost rather than lowering it. That is the trade a buyer actually signs up for: a smaller, cheaper detector paired with a faster, more tightly toleranced and more expensive lens.

Resolution requirements are splitting along the same fault line. Defence and autonomous driving are racing toward megapixel-class infrared; consumer electronics wants the cheapest possible sensor. VGA sits in the middle — convenient in size, awkward in position, because it is squeezed between the two trends rather than serving either of them cleanly.

Germanium is losing ground to chalcogenide glass

On the optics side the report identifies a materials migration: the industry is working to move from germanium to chalcogenide glass, with durability as the stated motivation. Germanium remains a workhorse for MWIR and LWIR lenses, but it is heavy, supply-sensitive and expensive to own through a product’s life, while mouldable chalcogenide glass supports higher-volume forming, wider operating conditions and better resistance to real-world handling and environmental exposure.

The same paragraph flags a second front: anti-reflective metasurface technology has started to enter commercial application. For infrared systems this matters because every interface in an LWIR lens train loses transmission, and AR metasurfaces are a route to higher throughput without stacking additional conventional coating layers.

At the low end, pyroelectric and thermopile devices occupy the low-cost, low-resolution band. Thermopile detectors have already penetrated smartphones, smartwatches, home appliances and electric vehicles, and miniaturisation is driving them into smart rings, smart glasses and true-wireless earbuds through package-level integration and SWaP-optimised modules. The report also notes that AI-enhanced thermopile arrays are beginning to compete with entry-level microbolometers, using blob-based sensing that stays privacy-preserving because it never resolves an identifiable image.

Infrared Lenses & Assemblies Sourcing →

Supply chains are being rebuilt along national lines

The report also describes a supply map in motion. The United States is working to build a self-controlled “sensors and optical components” supply chain, meaning US programmes drawing on US-made components. Europe is pursuing something similar in optical components, though with weaker regulatory protection than the US version. Meanwhile, despite new export restrictions cutting both ways, Chinese infrared detector makers continue to expand production capacity.

Who is shipping today is not in doubt in the report: Raytron (睿创微纳), HIKMICRO (海康微影) and Guide Infrared (高德红外) lead on unit volume. Western suppliers are positioned at the mid-to-high end and are increasingly bundling AI and rich software features onto the sensor, competing on capability rather than on cost per pixel.

What this changes for infrared optics buyers

• Fix the pixel pitch before you quote a lens. An 8 µm detector and a 12 µm detector of the same format do not share an optical design. The report’s point that smaller pixels demand larger apertures and tighter tolerances means the lens cost curve runs opposite to the sensor cost curve — model them together.
• Ask suppliers for a chalcogenide route as well as a germanium route. The migration to moulded chalcogenide is a durability and volume play. A supplier that can only grind germanium limits both your unit economics and your environmental specifications.
• Treat coating throughput as a line item. With AR metasurfaces entering commercial use, the coating and surface-relief choice is now a transmission budget decision, not a finishing step. Specify transmission targets per element and per band, and require measured curves.
• Do not plan a 2026 automotive IR volume assumption. The report explicitly describes automotive as deferred, not cancelled. If your business case depends on automotive infrared volume in the near term, it is early.
• Decide whether VGA is good enough, and say so in writing. VGA is the format the market can build cheaply at scale, but it sits between the megapixel defence push and the low-cost consumer pull. If your application can live with VGA, fix it early; if it cannot, the optical design gets materially harder.
• Match the sensing modality to the privacy and cost requirement. Blob-based thermopile sensing with AI is explicitly positioned against entry-level microbolometers. For occupancy, presence and gesture functions, that is often the cheaper and more privacy-acceptable answer than a full thermal image.

Sources and disclosure

Sources: MEMS Consulting (麦妆姓咨询), 28 September 2026, summarising the 2026 edition of Yole Group’s Thermal Imaging and Sensing report — defence (UAV, counter-UAS, border surveillance) as the primary growth driver for microbolometer supply, automotive potential deferred, industrial/consumer/medical steady, China’s 8 µm pixel-pitch push against the Western 12 µm standard and the larger-aperture, tighter-tolerance optics that smaller pixels require, VGA as the mid-format compromise, megapixel demand in defence and autonomous driving, the migration from germanium to chalcogenide glass for durability, anti-reflective metasurfaces entering commercial application, pyroelectric and thermopile devices in the low-cost low-resolution band with thermopiles already in smartphones, smartwatches, appliances and EVs, miniaturisation into smart rings, smart glasses and TWS earbuds, AI-enhanced thermopile arrays competing with entry-level microbolometers via blob-based privacy-preserving sensing, US and European moves toward domestically controlled sensors and optical component supply chains with weaker European regulatory protection, continued Chinese capacity expansion despite export restrictions, and Raytron, HIKMICRO and Guide Infrared leading shipments while Western suppliers target the mid-to-high end with AI and software integration. Procurement, qualification and supply-chain observations are TUOLAN TRADE analysis and are not attributed to any cited source. Edited and analysed by TUOLAN TRADE.