Im Jahr 2025 sind die KI-Brillen von ihrem Pedestal gestiegen.

AI glasses are experiencing a fate that fluctuates like a roller - coaster ride.

Data from IDC shows that in the second quarter of 2025, the global shipment volume of the smart glasses market reached 25.55 million units, representing a year - on - year increase of 54.9%. The Chinese market performed even better: the shipment volume amounted to 664,000 units, corresponding to an increase of 145.5% compared to the previous year. Many manufacturers have entered this market. From Xiaomi to Alibaba, from Rokid to Thunderbird - almost all relevant companies have presented plans for smart glasses this year.

In the first half of this year, the sales of smart glasses on the JD platform increased more than tenfold compared to the previous year. The number of registered brands has almost tripled. On average, a new product appears on the market every nine days.

But behind the guise of success lies a crisis. According to statistics from industry organizations, the return rate of AI glasses on platforms such as JD and Tmall is about 30%. On the Douyin platform, it is even between 40% and 50%. Behind the high return rate lies the problem of lack of functional usefulness, which consumers have cited as a central issue.

The once - highly - regarded AI glasses are now being questioned like never before.

On the Eve of Deflation

Xiaomi's AI glasses are almost a typical example of this bubble phenomenon. In the first week after a successful sale, 70,000 pairs were sold. According to official information, the return rate of these glasses on the Douyin platform is about 40%.

It is known that Xiaomi's AI glasses only took one year from concept to market launch. But this "speed" has led to a compromise in user experience and a high return rate. The reason is that the product is more of a simple combination of "headset + camera" and lacks innovative solutions to meet user needs.

A customer decided to return the glasses after five days of use. He admitted: "The Xiaomi glasses with only one camera and no lens display are more of a simple combination of 'headset + camera'. The actual user experience did not meet expectations." For example, in navigation, due to the lack of visual display and only voice output, it is difficult to find the right way at intersections. In the end, one still has to take out the smartphone to look at the map.

A technology blogger found during a test that one can successfully cover the recording warning light of Xiaomi's AI glasses with just a felt - tip pen. It is hardly noticeable from the outside whether the warning light is on. After the recording function is activated, people nearby hardly notice that the glasses are in recording mode. This has raised serious concerns about privacy.

Of course, this is not an isolated case. The highly - anticipated "battle of a hundred glasses" in 2025 did not break out as planned. The lack of industry standards is a persistent problem.

"So far, no manufacturer has covered all the steps from materials, optics, hardware, software, systems to algorithm development for AI glasses," said Cui Haitao, Chairman of Gudong Intelligence, who sums up the industry situation precisely.

The dreams are big, but the reality is harsh. According to insiders, in the early stage of the industry, players often took Meta Ray - Ban as an example. Product design was often influenced by MetaBan, but the focus was more on quickly entering the market. However, the actual products have fallen far short of users' expectations.

"The core of AI glasses lies in the ability to provide in - depth service value in the form of 'data accumulation + decision - making support' based on the technology of 'all - day wearability + visual and auditory perception'." However, although many players from different sectors - from AI algorithm companies, traditional glasses manufacturers, mobile phone manufacturers, audio devices, Internet conglomerates to telecommunications providers - have entered this market, they tend to stack hardware products rather than implement revolutionary innovations in core functions.

Li Chuangqi, head of Xiaomi's Wearable Division, admitted that Xiaomi's AI glasses use almost the most expensive components. Although the price of 1,999 yuan leaves a profit margin, considering the overall investment, it is still a loss - making business.

Players with different backgrounds are flocking to this market, each with their own calculations.

Internet giants such as Alibaba have introduced Quark AI glasses and are trying to "bring the entire Alibaba ecosystem to users' faces" - by integrating functions such as the Tongyi Qianwen large - model, Gaode Navigation, Alipay payment, and Taobao Picture Search.

After a series of organizational structure adjustments since the end of last year, the AI application "Tongyi", which was originally part of Alibaba Cloud, was integrated into the intelligent information business group, which also includes Quark. At the same time, the integration of Tmall Genie and Quark products is being promoted to explore new hardware directions such as AI glasses. This is coordinated by Wu Jia, CEO of Alibaba's intelligent information business group.

According to media reports, the standard price of Quark AI glasses of 2,999 yuan corresponds to the cost. If one does not make a profit from the products themselves, the public speculates that Alibaba's ambition is aimed at charging for ecosystem functions and scenarios.

Mobile phone manufacturers such as Xiaomi regard AI glasses as part of their strategy for the "entire ecosystem of cars, mobile phones, and homes". They hope that together with mobile phones and cars, they will form the three core end - devices of the Xiaomi ecosystem.

Companies such as Rokid and Thunderbird have decided to specialize in specific areas. Zhao Weiqi, head of Rokid's global developer ecosystem, said: "Currently, the AI glasses market is still in a healthy competition."

But since there are still no unified standards and regulations in the industry, "discussions and research on regulating AI use and avoiding privacy violations are still in the early stage," said Wang Yiru, financing director of XREAL. Currently, one can only predict some risks and has to solve problems during the development process.

The technological development in the AI glasses industry is accelerating. Brands must take advantage of the opportunities in the supply chain and ecosystem cooperation. However, in reality, the industry is in a vicious circle of "high investment, low returns, and difficult implementation".

Why are AI glasses developing so slowly?

Rokid Glasses had an impressive debut at the World Artificial Intelligence Congress this year and attracted many visitors to line up for the testing phase. But consumers quickly found that there is a big difference between theory and practice - the glasses try to replace the smartphone, but the actual user experience cannot do without the smartphone.

Capacity is also a big problem. Many brands have presented impressive demos, but the actual supply capacity lags far behind. Some consumers said that they waited for several months after pre - ordering Rokid glasses and that the capacity is still limited. Finally, in September, some products were launched on the market, but the price has risen from the reserved 2,499 yuan to 3,299 yuan.

Industry insiders have admitted the reality. Consumers expect AI glasses to have functions such as multimodal recognition, calls, video recording, and live - streaming, and at the same time, they can be worn stylishly and unobtrusively. The requirements for manufacturers are very high, but in reality, it is difficult to meet these expectations with the existing technology.

This is directly related to the immaturity of the AI glasses supply chain. Since the sales volume of AI glasses is still low and the user experience is not satisfactory, the corresponding chips, edge - AI algorithms, new materials, etc. are not yet mature. Manufacturers also do not have a separate production line for AI glasses in mechanical manufacturing.

The industry generally faces the "impossible triangle" situation - performance, lightness, and battery life are difficult to reconcile.

Greater computing power requires a chip with higher power consumption, which shortens the battery life. A longer battery life requires a larger battery, which increases the weight. Lower weight is a prerequisite for all - day wearability, but it restricts the battery capacity and heat - dissipation space.

So far, no manufacturer has covered all the steps from materials, optics, hardware, software, systems to algorithm development for AI glasses.

In terms of algorithms, there are no special AI algorithms for glasses. Therefore, AI glasses have to use small models on the mobile phone side, which makes integration difficult. For optical modules, currently, medium - sized frames are supported, but not large or small frames.

Whether it is Meta or Chinese manufacturers, most use "general" solutions such as the Qualcomm Snapdragon AR series. These chips are essentially variants of mobile phone SoCs and are not developed for the extreme scenarios of AI glasses - "all - day use, minimal power consumption".

An industry insider said quite frankly that the entire industry is waiting for a special chip that can handle complex AI tasks with a power consumption in the milliwatt range. Without this "heart", AI glasses can never get rid of the embarrassing image of a "power - bank accessory".

A deeper problem is that AI glasses lack a real "killer app". Most of the current functions such as translation, navigation, photography, etc. can be better done by smartphones. AI glasses try to combine "headset + camera + information assistant" in a single device, but for each individual function, they cannot offer a better experience than established devices.

A technology blogger described the experience of the image recognition function of AI glasses: One has to take a photo first and send it to the AI. After waiting for a few seconds to a few tens of seconds, the glasses can return the result. This experience is very different from the "always - at - hand AI question" promoted by manufacturers.

Battery life is also a big problem. A customer complained: "The recording time is clearly too short. The continuous use time is only about 2 - 3 hours."

In the truest sense of the word, AI glasses are still new, but this industry has already undergone a restructuring. In comparison, some products that focus on optimizing existing functions, such as AI speech recorders, have received a good response in the market.

The reason behind this contrast is that the latter meet clear needs and offer a really useful user experience, rather than just talking about concepts.

The Dreams and Reality of AI Glasses

The problem that AI glasses face essentially lies in the discrepancy between supply - chain capabilities and market needs.

On the one hand, the functions marketed by manufacturers are not really needed by consumers, while the experiences desired by consumers cannot be realized due to technological limitations. On the other hand, manufacturers paint dreams of a "next - generation computing platform", while consumers are increasingly emphasizing that they need practical tools to solve their problems.

According to market feedback, AR glasses with displays dominate the market. According to statistics from CINNOResearch, the sales of the Chinese consumer market for AI/AR glasses in the first quarter of 2025 increased by 45% compared to the previous year.

AR glasses with displays account for 80% of the market, while AI glasses without displays only account for 20%. This shows that users' requirements for the glasses - the display is far more important than the AI function.

Optical modules account for up to 40% of the total cost of the device. The common solutions with MicroOLED, MicroLED + diffractive waveguide technology are expensive, the production technology is complex, and the yield is low. Therefore, the prices of consumer models have remained above 3,000 yuan for a long time.

This has prevented most manufacturers from going beyond small - scale tests.

The high costs not only restrict mass production but also keep the user group limited to geeks, niche creators and other circles for a long time, preventing mass dissemination.

At the same time, the sales channel is also suffering. When traditional glasses stores sell smart glasses, they have to face difficulties...