Three Major Trends of Infrared Optics Industry in 2026: Miniaturization, AI Integration And Multi-Spectrum Technology

Brief Introduction

With mass commercialization of L2+ and L3 autonomous driving, vehicle-mounted infrared lenses have evolved from optional night vision accessories for premium vehicles into irreplaceable redundant sensors of multi-sensor perception systems. In 2026, China’s infrared optical industrial chain realized full domestic substitution of chalcogenide glass, vehicle-grade infrared detectors and dedicated ISP chips. The technological development of the whole industry converges into three clear directions: hardware miniaturization, deep integration of thermal imaging and edge AI, and integrated multi-spectrum optical design. Based on public industrial data from ZuoSi Automotive Research, Yole Développement and MEMS Consulting, as well as mass production solutions from leading manufacturers including Sunny Optical and Raytron Technology, this paper analyzes the technical logic, mass production status and market value of the three trends respectively.

Here is Contents:

1.Miniaturization: Chalcogenide Molding & Small-Pixel Packaging for Hidden Vehicle Layout

2.AI Integration: Built-In Edge Computing on Lens Modules for Local Thermal Image Preprocessing

3.Integrated Multi-Spectrum Technology: Co-Aperture Dual-Spectrum Lenses Become Standard ADAS Hardware

4.Synergistic Industrial Value and Short-Term Application Limitations of Three Trends

1. Miniaturization: Chalcogenide Molding & Small-Pixel Packaging for Hidden Vehicle Layout

The miniaturization of infrared lenses in 2026 is driven by two major technological breakthroughs in optical materials and detector packaging, solving long-standing pain points of traditional germanium lenses such as large volume, heavy weight and limited assembly space. In terms of materials, precision compression molding for chalcogenide glass has achieved stable mass production. Compared with lenses cut from monocrystalline germanium, chalcogenide infrared lenses are 35% thinner and 60% lighter, while production costs are reduced by more than 70%. Domestic optical suppliers including Sunny Optical and Phoenix Optical have mass-produced ultra-short focal length infrared lenses with 3.1mm focal length and F1.0 large aperture, which can be hidden inside vehicle bumpers and grilles without separate external openings, matching the integrated design trend of new energy vehicles. For detectors, 8μm and 12μm small-pixel wafer-level packaging (WLP) products have passed vehicle-grade certification and entered mass supply. The SWLP packaged infrared detector modules launched by Raytron Technology are 40% smaller in overall size, and the total thickness of complete infrared camera modules is controlled within 20mm, greatly reducing layout difficulty for automakers. Industrial statistics verify this trend: the average focal length of vehicle infrared lenses delivered in 2025 was 80mm, while mainstream focal lengths dropped to 35–50mm for new vehicles in 2026. The penetration rate of miniaturized infrared lenses on mid-range vehicles below 300,000 RMB rose from 4.1% in 2025 to 12.5% in 2026. Miniaturization effectively cuts supporting costs and appearance design restrictions, accelerating the popularization of infrared night vision from luxury SUVs to mass-market new energy vehicles.

2. AI Integration: Built-In Edge Computing on Lens Modules for Local Thermal Image Preprocessing

The most transformative shift of the industry in 2026 lies in the repositioning of infrared hardware: from simple image acquisition optical components to intelligent perception units integrated with edge computing power. AI algorithms are embedded into built-in ISP chips inside lens modules, replacing the old mode that all raw gray thermal images are transmitted to vehicle domain controllers for centralized processing. On the hardware side, domestically developed vehicle-specific ASIC-ISP chips for infrared imaging have entered mass commercial supply. Infrared lens modules integrate lightweight edge computing units to locally execute thermal image super-resolution restoration, heat source segmentation and thermal clutter suppression on hot summer road surfaces, without transmitting massive raw images. This design cuts transmission bandwidth by 60% and shortens perception delay to less than 15ms, guaranteeing real-time response of AEB automatic emergency braking. In practical applications, shutter-free AI calibration and Matrix IV thermal image super-resolution algorithms become standard module configurations. After AI enhancement, the recognition accuracy of pedestrians and two-wheelers based on 640×512 thermal imaging reaches 94.6%, far exceeding traditional pure image processing schemes. Mass-produced intelligent driving solutions including Great Wall Tank series and Huawei ADS 3.0 adopt built-in lens AI to distinguish human and animal heat sources at pixel level, reducing the missing detection rate of night crossing pedestrians by 78%. From the perspective of industrial competition pattern, the core competition among optical suppliers in 2026 has shifted from optical parameter comparison to complete integrated hardware & algorithm solutions. The bargaining power of manufacturers only supplying optical lenses keeps declining, while orders for integrated modules with local AI processing account for over 65% of total orders.

3. Integrated Multi-Spectrum Technology: Co-Aperture Dual-Spectrum Lenses Become Standard ADAS Hardware

Single long-wave infrared or visible light lenses have inherent perception defects. In 2026, multi-spectrum fusion technology has completed upgrading, advancing from backend algorithm fusion to front-end hardware co-aperture integration. Dual-spectrum lenses combining infrared and visible light become mainstream hardware solutions for high-level ADAS. Traditional split dual-camera schemes require two independent optical paths and two body openings, leading to high assembly costs and natural parallax deviation between two groups of images, increasing post-processing algorithm registration workload. New-generation co-aperture multi-spectrum lenses adopt beam splitter prism structure; one set of chalcogenide optical path synchronously outputs RGB visible color images and 8–14μm long-wave thermal images with natural pixel alignment, eliminating extra image registration algorithms and cutting overall vehicle assembly cost by 28%. The two spectrums form complementary perception advantages: visible light identifies lane lines and traffic sign texture details, while infrared stably detects living targets such as pedestrians at night, in heavy fog and under strong glare. After fusion, the obstacle detection distance doubles under foggy conditions with visibility of 50 meters. According to research data from MEMS Consulting, the penetration rate of dual-spectrum infrared lenses on domestic mass-produced passenger vehicles will reach 18.3% in 2026, and exceed 45% by 2030 with a 32.7% compound annual growth rate. Besides exterior long-wave dual-spectrum lenses, integrated cabin lenses combining 940nm near-infrared and visible light also achieve rapid volume growth, complying with global mandatory DMS driver monitoring regulations effective in 2027 to identify driver fatigue and distraction all day long. The industry forms a dual supporting layout: exterior long-wave multi-spectrum perception for environment detection and interior near-infrared multi-spectrum monitoring for cabin safety.

4. Synergistic Industrial Value and Short-Term Application Limitations of Three Trends

Synergistic Industrial Value

Miniaturization breaks installation space bottlenecks for vehicle equipment; built-in edge AI reduces computing pressure on vehicle domain controllers; integrated multi-spectrum technology covers perception defects of single sensors. The three trends jointly form a complete low-cost, high-reliability all-weather perception solution. Multiple institutions estimate that domestic vehicle infrared lens shipments will exceed 10.71 million sets in 2026, with the corresponding market scale reaching 1.93 billion RMB, a year-on-year increase of 22.7%.

Short-Term Application Limitations

First, the precision beam-splitting coating process required by co-aperture multi-spectrum lenses has high technical barriers, and the number of mature domestic mass production lines is insufficient, resulting in higher short-term supply prices than split dual-camera schemes. Second, the mass production yield of 8μm ultra-small pixel detectors remains insufficient, keeping the overall cost of high-resolution miniaturized infrared modules at a relatively high level. Third, infrared AI algorithms still need further iteration to suppress thermal clutter from high-temperature summer road surfaces, and the problem of false recognition of small obstacles under complex driving scenarios remains to be improved.

Conclusion

In 2026, the infrared optics industry enters a coordinated upgrading stage driven by three core trends: miniaturization, AI integration and multi-spectrum integration. The three technical directions break original bottlenecks of vehicle-mounted infrared lenses in hardware size, intelligent processing capacity and multi-dimensional spectral information respectively. Chalcogenide miniaturized optical paths cut vehicle assembly space and hardware costs; built-in edge AI inside lenses improves real-time perception performance; co-aperture multi-spectrum hardware realizes all-weather full-scenario sensor fusion. Although short-term process constraints such as coating yield and detector production restrict full-scale industry popularization, continuous technological breakthroughs in domestic optical manufacturing, infrared chips and intelligent recognition algorithms will accelerate the standardization of the three technical routes. Infrared lenses will gradually become standard perception hardware for L2+ and above ADAS systems, continuously releasing long-term growth potential of the vehicle infrared optics track.