The Price Dilemma of Micro-Optical Engines for AR Glasses, Seen Through the Supply Chain

Today, many consumers have already seen the dazzling footage of AR glasses in short videos: put on the device and virtual images float before your eyes, with navigation and messages overlaid directly onto the real-world view.

Yet most mass-produced models still cost several thousand yuan or even more than 10,000 yuan. Industry insiders know that the core bottleneck keeping AR devices out of mainstream pockets is not the chip or the software, but a palm-sized component hidden inside the temple: the micro optical engine.

The micro optical engine is the image-generation core of AR glasses, essentially a miniaturized version of a traditional display that projects images onto the retina. Capital continues to bet on AR, while large AI models and spatial interaction features keep evolving, but product rollout remains stalled by hardware costs.

Beyond cost, the biggest challenge in the micro optical engine field today is that laboratory prototypes deliver stunning results, but once production scales to millions of units, cost and yield problems erupt all at once.

Why is that?

01  The "small player" inside AR glasses

At first glance at AR glasses, you cannot see the micro optical engine. Yet every image it displays depends on it.

Put simply, a micro optical engine is a highly compact projection system that packs the image light source, optics lenses and driving components into an extremely small space and projects the image into the human eye, much like the cornea, which responds to and processes the light entering the eye. At the same time, its weight and size must not burden the wearer.

The mainstream approaches on the market fall into three categories: LCoS, Micro OLED and DLP. Micro OLED and Micro LED are the two dominant display technologies in consumer AR today, with Micro LED holding greater market appeal thanks to its brightness advantage.

According to estimates in a Great Wall Securities industry weekly, integrated AR glasses will account for 18% of online smart glasses sales in China in 2024, rising to a range of about 22%-27% in 2025. The micro optical engines used in many mainstream models are assembled from more than ten precision parts, including display chips, collimating optics and polarizing films. Assembly tolerances must be held to the micron level, equivalent to a fraction of the diameter of a human hair.

This explains why laboratory prototypes can be impressive but fall short after mass production. At the prototype stage, manufacturers need to build only tens or hundreds of units, and engineers can hand-tune each lens assembly, making it easy to hit the target image quality. Once mass production begins, even tiny assembly deviations directly cause blurry images, ghosting and uneven color, and the product is rejected outright.

In fact, most AR brands cannot buy mature optical engines off the shelf; they must either develop them in-house or place custom orders with a small number of optical solution providers. The mature supply chain division of labor seen in other consumer electronics categories has yet to be established in the micro optical engine segment. AR device makers can easily purchase mature general-purpose chips, but micro optical engines are customized components with no standardized, off-the-shelf products available.

The tougher cost challenge lies deeper in the supply chain and manufacturing.

02 Why are lab samples cheap, but prices rise in mass production?

Currently, cost pressure on micro-optomechanical systems stems from constraints such as yield, scale, and components.

Production yield is a critical barrier. Because optical assembly is ultra-precision manufacturing, dust, temperature fluctuations and tiny errors in assembly tooling can all cause optomechanical failure. In the early mass-production stage, yields in optical coupling and lens processing were relatively low, and some processes had high scrap rates.

On production lines, large numbers of products are scrapped outright during inspection. The raw material and labor costs of those rejects are all allocated to the units that pass, directly driving up the selling price per machine.

Another awkward issue is that the promised economies of scale have been slow to materialize.

IDC's 2025 China Smart Glasses Market Tracker report shows that domestic AR/VR device shipments totaled 742,000 units that year. Compared with the smartphone market's annual volume of well over a billion units, this market is far too small. Upstream component suppliers are reluctant to invest in dedicated production lines for expansion; without large-scale lines, unit component costs cannot come down; and high-priced finished products deter ordinary consumers, making a sales surge difficult and trapping the industry in a classic chicken-and-egg cycle.

At the component level, core optical films and microdisplay chips are highly concentrated in the hands of a few overseas suppliers. According to relevant research, the microdisplay module (emitter chip plus driver circuit) is the most expensive part of the optical engine, accounting for about 50% of the finished product's BOM cost. Similarly, high-end optical lenses and polarizing films are also heavily dependent on imports.

High costs therefore hold back consumer AR adoption, and the supply chain is looking for viable ways to break through.

03  Is there a realistic path to solving mass-production costs?

Cost reduction for micro-opto-mechanical systems is a hurdle the entire AR industry cannot avoid. If the supply chain can overcome the twin barriers of yield and cost, AR glasses could follow smartphones on the path to mass adoption.

The prerequisite is solving mass-production cost issues with ease. Today, players in this space are compressing micro-opto-mechanical costs from multiple directions: domestic substitution, structural simplification, and process iteration. Each approach involves trade-offs, and no perfect solution exists.

Substituting domestic optical components is the most direct short-term measure. Chinese optical makers are gradually moving into the production of components such as polarizers and optical waveguide lenses, breaking the monopoly of overseas suppliers.

After core optical materials are replaced with domestic alternatives, optical waveguide costs fall by more than 60% versus imported solutions; localizing ordinary auxiliary optical films can cut costs by 15% to 20%, as confirmed by public technical disclosures from domestic solution providers such as Gudong Technology. However, the localization of high-end Micro LED chips is progressing slowly, and core display chips remain in short supply.

Simplifying the optical architecture can trade some performance for lower costs. Many AR products aimed at the mass market have abandoned complex optical waveguide solutions in favor of relatively simple optical architectures such as BirdBath. These designs deliver slightly lower image clarity, but use fewer parts, are easier to assemble, achieve higher yields more readily, and can substantially reduce the optical engine cost per unit.

Most lightweight AR glasses focused on video and entertainment currently follow this approach, prioritizing mainstream audiovisual needs. After all, consumer users care less about the technology you use than about the viewing experience.

In the AR/VR industry, full-color MicroLED optical engines have long been regarded as the core direction for next-generation display technology. Kangtaida, for example, has unveiled production equipment for full-color MicroLED optical engines. The key advantages of the system are said to be cost reduction and efficiency gains: modular assembly cuts the footprint by 22%, changeovers take 30 minutes, the full-color first-pass yield reaches 99% (excluding material-related causes), cycle time is under 60 seconds, and UPH rises by 100%. Overall, the equipment can reduce the cost of MicroLED optical engine products by 28% and shorten delivery cycles by 50%.

Clearly, the entire supply chain needs coordinated engineering. Optical materials, chips, precision assembly and inspection equipment: if any link falls behind, it will hold up cost reductions. Even if optical engine costs fall, weight, image quality and battery life must still be balanced; cutting costs alone can easily come at the expense of the wearing experience.

On the other hand, capital's vision for AR spatial computing mostly centers on AI interaction and future scenarios blending the virtual and real. Many investors focus on software capabilities such as spatial computing and large-model interaction, while overlooking the practical constraints of hardware manufacturing.

The history of consumer electronics has repeatedly shown that no product can truly enter ordinary people's lives without clearing the price barrier. Software algorithms can be iterated quickly and upgraded via the cloud, but micro optical engines must be refined step by step through factories, production lines and processes. There are no shortcuts.

By R Xingren

       Original title : Dead on Arrival at Mass Production? The Price Dilemma of Micro Optical Engines from a Supply Chain Perspective