In the first half of 2026, multiple AR optical waveguide companies are sprinting towards a capacity of "one million units": Zhige Technology aims to deliver 1 million Surface Relief Grating (SRG) optical waveguides in 2026; AAC Technologies has planned an annual SRG lens capacity of 1 million units; Nika Optics' automated production line for 1 million units of Volume Holographic Grating (VHG) optical waveguides has officially commenced production; and Raypai Crystal Optics expects to increase its annual capacity of geometric waveguide optical modules to 1 million sets after its new site is put into operation. Meanwhile, new optical waveguide technology routes are moving from the laboratory to the market, entering the initial stage of mass production. In December 2025, Parallel Vision Group's global first pilot production line for Polarization Volume Grating (PVG) optical waveguides was officially completed, with the capability to produce 50,000 optical waveguide wafers annually. Focusing on the technological iteration direction of optical waveguides for AI glasses, the mass production status of various technology routes, and the current pain points of the industry, reporters from China Electronics News interviewed industry insiders.
"Wearable" AI Glasses Propose Over a Dozen Metrics for Optical Waveguides In the technical solutions for AR glasses, optical waveguides have emerged as a rising star due to their advantage of reducing lens thickness to the sub-millimeter level. According to Counterpoint Research data, in the first quarter of 2026, global AR glasses shipments increased by 136% year-on-year. Among them, the market share of mature solutions such as Birdbath and FlatPrism dropped from 82% in the same period last year to 58%, while the market share of optical waveguide AR glasses increased from 18% to 42% over the same period.
This new terminal form of AI glasses is opening up broader market space for optical waveguides. Most traditional smart glasses follow a "display-free" route, achieving functions such as photography, translation, and Q&A through cameras, microphones, speakers, and basic motion sensors. However, for smart glasses to truly become "AI glasses" or even "agent glasses," they need to meet multimodal interaction requirements, and "vision" is an indispensable and even the most critical part of multimodality. More and more tech giants and AR unicorns, such as Meta, Alibaba, Rokid, and Even Realities, are shifting to the "AI+AR" route, enabling AI information to be projected onto the smart large screen constructed by AR.
So, what standards must be met to create an ideal pair of "AI+AR" glasses? The lighter the weight, the better; the thinner the lenses, the better; the larger the FOV (Field of View), the better; and they must be full-color, highly transparent, and cost-effective—in short, the closer the wearing experience is to ordinary glasses, the more likely consumers are to wear them and buy them.
However, when such requirements are implemented in optical waveguide solutions, they correspond to over a dozen or even more metrics.
Meng Xiangfeng, CEO of Beijing Zhige Technology Co., Ltd. (hereinafter referred to as "Zhige Technology"), told reporters from China Electronics News that the performance considerations of optical waveguides for AI glasses can be viewed from two levels.
The first level is the wearing experience.
On the one hand, the waveguide wafer must steadily reduce its weight and thickness in line with the downstream manufacturers' endless pursuit of thinness and lightness.
On the other hand, it is necessary to steadily improve and eliminate the current defects of grating technology. First, eliminate the rainbow effect. When external natural light passes through the diffraction grating, it is dispersed into rainbow-like stray light, which interferes with the consumer's field of vision, so this stray light interference must be eliminated. Second, solve the light leakage problem to prevent the images displayed by the waveguide wafer from being captured by the outside world and leaking user privacy. Third, improve the transmittance. Ideally, it should be like high-end myopia glasses, increasing the transmittance to over 98% to provide users with a bright field of vision. Fourth, reduce the "grating visibility" (referring to the color difference between the area of the lens with the grating structure and other areas). How to make the grating highly transparent so that consumers cannot feel it is also a challenge.
The second level is optical metrics.
First is the FOV, which downstream customers care about most; the larger the FOV, the stronger the sense of immersion. Second is optical efficiency (the efficiency of light energy utilization). If the optical efficiency is high enough, even with a lower light engine driving current, a high-brightness image can be produced, thereby improving battery life and reducing battery weight. Third is image uniformity. On the one hand, the image needs to be made uniform in optical design; on the other hand, it is necessary to follow the display industry and introduce correction algorithms for uniformity. Finally, there are detailed metrics such as contrast and clarity.
While pursuing ultimate optical performance, commercialization and engineering implementation are equally important, as they relate to the cost and capacity of the final product.
Zhang Yuning, CEO of Nanjing Parallel Vision Technology Co., Ltd. (hereinafter referred to as "Parallel Vision Group") and professor at Southeast University, pointed out to reporters from China Electronics News that AI glasses makers’ focus on the commercial viability of optical waveguides falls into three areas: first, the comprehensive module cost. The room for cost reduction of the waveguide module and related optical components determines whether the whole machine can enter the consumer-grade price range. Second, scaled capacity and delivery. The scaled mass production capability and yield stability of the production line are the guarantees for large-batch, high-consistency delivery. Third, environmental reliability, including consumer electronics-level reliability testing standards such as high and low-temperature resistance, drop resistance, and heat and humidity resistance.
In addition, downstream manufacturers highly value the evolution ceiling and bottlenecks of optical waveguide technology routes. "Downstream manufacturers will evaluate whether there are insurmountable physical or engineering bottlenecks in this technology route, such as too low an upper limit of optical efficiency, overly complex full-colorization, or lens thickening caused by stacking. They tend to choose directions that combine current availability and future upgrade space," Zhang Yuning said.
Multiple Technology Routes Compete, Stepping on the Accelerator for Optical Waveguide Mass Production Facing the market opportunities brought by AI glasses, the mainstream technology routes of optical waveguides have started a mass production race. Currently, the technology routes for AR glasses optical waveguides are mainly divided into geometric waveguides, Surface Relief Grating (SRG), Volume Holographic Grating (VHG), and Polarization Volume Grating (PVG). In terms of physical mechanisms, geometric waveguides use geometric optics, while SRG, VHG, and PVG use diffractive optics.
In the diffractive optics camp, SRG dominates in market share, industrial maturity, and customer acceptance. Meng Xiangfeng told reporters that the advantage of SRG lies in its ability to rely on semiconductor processes. Semiconductor leading enterprises have entered the fields of materials, processes, and equipment, forming a "flywheel effect" contributed by many entities across the industry. However, its yield improvement also faces two major challenges: first, appearance quality. Whether it is the impact of particulates or scratches, it will cause lens defects, which will be illuminated and amplified during the interaction between light and the waveguide wafer. Second, optical consistency. In the continuous production of tens of thousands or even hundreds of thousands of lenses, it is necessary to ensure that the optical performance (such as optical efficiency, color uniformity, and other optical metrics) of each piece is within the tolerance range of the design target.
As early as 2023, Zhige Technology built a fully automated batch production line for diffractive optical waveguides, aiming to deliver 1 million optical waveguides in 2026. Meng Xiangfeng stated that currently, the first-pass yield of Zhige Technology has reached over 85%, with some projects exceeding 90%. "In the process of developing new materials and new processes, we have been promoting suppliers to make progress together. Compared with the standard machines of equipment manufacturers, our customized machines and developed processes can increase the yield by dozens of percentage points, which also drives the technological progress of the industry and the performance growth of suppliers. We hope to continue refining our technology, take a leading role within the supply chain and drive industry progress," Meng Xiangfeng said.
Meanwhile, "next-generation volume holography" technologies, including HPDLC (Holographic Polymer Dispersed Liquid Crystal) and PVG, are emerging as a powerful force. Zhang Yuning told reporters that the PVG route not only inherits the inherent advantages of volume holography, such as high optical efficiency and low light leakage, but also opens up the evolution channel for a large FOV through the molecular self-assembly response characteristics of polarized liquid crystals. Moreover, it has the potential for automated large-area production and comprehensive cost reduction, showing broad prospects and advantages.
At the engineering level, there are two core difficulties in PVG mass production. First, there is a lack of ready-made equipment and experience to follow. Currently, there is no large-area, high-precision mass production equipment specifically for the special optical structure and material system of PVG on the market. Second, there is a lack of "full-chain deep coupling." The scaled mass production of PVG cannot be achieved by directly purchasing standardized equipment. It must start from the underlying photo-alignment and liquid crystal material R&D, reverse-engineer and explore the matching process window, then redefine and customize non-standard automated equipment based on process requirements, and finally continuously calibrate equipment and process parameters in actual production to run through stability and improve yield.
"This requires running through the entire process from underlying materials, optical processes to equipment control for systematic synchronous development. This also requires the R&D team to understand both underlying materials and optical processes and equipment customization, possessing full-chain engineering implementation capabilities," Zhang Yuning pointed out to reporters.
In December 2025, Parallel Vision Group's pilot production line for Polarization Volume Grating (PVG) optical waveguides was officially completed in Ji'an City, with the capability to produce 50,000 optical waveguide wafers annually. Zhang Yuning stated that Parallel Vision Group mainly paves its path to mass production through "source-level innovation from university research teams" and "engineering delivery via deep industry–academia–research integration." First, it leverages the underlying R&D heritage of university research teams to overcome the molecular design, phase transition control, and reliability challenges of polarized liquid crystal materials from the source, providing underlying theoretical and material support for mass production. Second, it connects the entire chain of "academic breakthrough—engineering transformation—equipment customization." Based on the independently developed material system, it carries out in-depth cooperation with upstream equipment manufacturers to jointly define and customize dedicated automated coating and exposure curing equipment. Third, it uses the pilot production line to run through industrial yield, continuously accumulates production line data, and optimizes fitting parameters, gradually refining a standardized production process with high yield and high consistency, laying the foundation for subsequent larger-scale commercial mass production.
The geometric waveguide, which does not rely on gratings, has high optical efficiency, good display color, and FOV. However, in terms of processing technology, it requires grinding and coating multiple layers of lenses at an oblique angle and stacking them together, making mass production difficult. Raypai Crystal Optics improves the yield and mass production feasibility of geometric waveguides through paths such as "2D pupil expansion" technology and self-developed molecular bonding processes. In May 2026, Raypai Crystal Optics announced the comprehensive deployment of a new generation of intelligent optical production lines in the Pingqian Maigao Park. After the new site is put into operation, its annual capacity for optical modules will increase from 100,000 sets to 1 million sets.
Avoiding Low-Quality "Involution", Breaking Through with Original Technology and User Experience Although AR optical waveguides have opened the window for mass production, capacity figures do not represent the maturity and health of the industry. Original technological breakthroughs and practical and feasible R&D and production goals are more critical to the overall progress of the industry.
Currently, the industrialization of AR optical waveguides still faces problems at three levels.
At the hardware level, the full-color AR industry chain needs to be improved. Zhang Yuning pointed out that, on the one hand, the yield, optical efficiency, resolution, and color performance of upstream micro-display sources/light engines (such as full-color Micro-LEDs) still need to be broken through; on the other hand, the cost of achieving full-colorization for optical waveguides is too high, and there are still objective bottlenecks in performance such as image uniformity and color deviation.
At the process level, there are still multiple technologies to be conquered and multiple routes to be verified. Meng Xiangfeng mentioned that the current mass production technology for nanoimprint lithography is relatively mature, but the mass production of 12-inch wafers still needs to be conquered. At the same time, the industry is also introducing semiconductor etching processes, combining materials with higher refractive indices to promote the high-end development of waveguide wafers.
At the ecosystem level, the content and scenario ecosystems are not yet ready. Although large models have endowed glasses with stronger perception capabilities, full-color display still needs to find irreplaceable interactive value and rigid user demands. Application scenarios and content ecosystems are still in the exploration stage.
To promote the high-end development of AR optical waveguide wafers, accelerate the process of commercial landing, and expand market size, it is necessary to be grounded in user experience and dare to act in original technology.
Focusing on user experience, Zhang Yuning suggested pragmatically promoting monochrome waveguide wafers in the short term, first making the "wearing experience" ultimate. "The industry does not need to blindly leap stages. In the short term, making monochrome optical waveguides ultimate in high transmittance, low power consumption, extreme thinness and lightness, and imperceptible wearing to meet the needs of truly low-cost information visualization functional scenarios, allowing users to be able and willing to 'wear them for a long time,' is itself a very critical leap," he said.
Regarding original technology, Zhang Yuning mentioned that it is necessary to adhere to underlying technological innovation and deep synergy between industry, academia, and research, base on independent intellectual property rights, and promote the domestic breakthrough of key materials and dedicated automated equipment. At the same time, avoid homogeneous cutthroat price competition and steadily improve mass-production yield.
Meng Xiangfeng also emphasized that the optical waveguide industry has two core barriers to entry: optical design and semiconductor manufacturing. There is a phenomenon in the current industry of emphasizing capacity over R&D, neglecting R&D investment in optical design, and even blindly imitating existing products and lowering prices, which may lead the industry into "involution." "I hope that practitioners realize that our industry is driven by the dual wheels of optical design and semiconductor manufacturing. If we emphasize capacity over R&D, it will hurt the enthusiasm of R&D personnel and lead the industry into low-level redundant competition," Meng Xiangfeng appealed.
Author | Zhang Xinyi Editor | Qiu Jiangyong Art Editor | Malia Supervisor | Lian Xiaodong