GuShiio.com Wen| Market Value List, Author| Xiao Rong,edit| Jia Lele
2025 is recognized as the first year of the explosion of AI glasses. But from another perspective, it is also the year when AI glasses are built.
From the CES exhibition at the beginning of the year to the upcoming Xiaomi AI glasses, this year, with the support of artificial intelligence, AI glasses are destined to become the next attraction for major Internet companies.
In January, Thunderbird Innovation released an AI shooting glasses called Thunderbird V3 at CES, equipped with a customized large model of Alitong Yiqian.
Coincidentally, the domestic smart AR glasses company Rokid has also launched a new Rokid Glasses AI glasses relying on the Alitong Yiqian Model, focusing on intelligent voice interaction and real-time information query.
In addition, many domestic manufacturers such as Li Weike, Mojie Technology, and Shunwei Technology also demonstrated their own AI glasses products or solutions.
When Apple Vision Pro came out two years ago, people still felt the regret that Cook failed to make AI glasses in one step.
To this day, Cook has felt the fierce competition in the glasses market.
On January 30, Zuckerberg said on Meta’s earnings conference call that Meta has high hopes for Ray-Ban Meta AI glasses and believes that it may be the next major area, and hopes to determine its market potential this year.
Domestic manufacturers are also coming in one after another. Baidu has released native AI glasses equipped with the Wenxin Yiyan model in November 2024, and Xiaomi’s AI glasses are getting closer and closer to official release. Huawei is also relying on its own big model to try to improve and update its original smart glasses functions.
But to this day, AI glasses are still a niche thing on the market.
The Northeast Securities Research Report shows that in 2024, the domestic sales of AI glasses will be approximately 50,000 pairs, accounting for 3% of the global total. Therefore, this blue ocean looks very broad. There are many views that 2025 is the first year of the outbreak of AI glasses.
But the road to AI glasses is obviously not an overnight move. It is necessary not only to positively view future outbreaks, but also to face current problems solidly.
Why are major manufacturers breaking into this seemingly small market? Will AI glasses usher in a major explosion due to the entry of major manufacturers? Who will break through in the fierce competition for AI glasses? This article attempts to answer these questions.
1. Why did big factories break in one after another
More than a decade after Google Project Glass fell from the altar, with the support of the big model, a new generation of AI glasses is following the same logic and reshaping the ecological landscape of the Internet giant.
First, the entrance is still the most direct and important value of AI glasses to large Internet companies.
If we want to transfer the Internet ecosystem to everyone’s life, a hardware device as an interactive portal is indispensable.
When smartphones have entered toothpaste squeeze innovation and new energy vehicles have fallen into a tragic knockout round, a new entrance is enough to cheer up all major Internet companies trapped in the era of stock again.
This is not just adding a connected device to the ecosystem, but because the connection of new devices to existing ecological devices will have the opportunity to bring new traffic increments to Internet companies. When users wearing AI glasses use their glasses to operate surrounding hardware and transmit images, it will also add many new variables to their lifestyle.
More importantly, the entrance effect has given big Internet companies the opportunity to tell new stories, and AI glasses, like previous smartphones and new energy vehicles, are the best story material.
Second, when today’s competition among major Internet manufacturers is reflected in a battle between hundreds of models, the story of smart glasses lies not only in the connection and linkage with existing equipment, but also in finding new application opportunities for large models.
Just as Zhang Yong, former CEO of Ali, proposed that all products are worth redoing with AI, the implementation of artificial intelligence technology is first reflected in the native application of software.
In 2023, Microsoft launched Copilot and applied it to Office. Since then, conversational software such as ByteDance Bean Bag and Kimi on the Dark Side of the Moon have followed the footsteps of ChatGPT in people’s lives.
AI glasses made the entire Internet circle notice for the first time the possibility of end-side applications of large models.
In November 2024, when Xiaodu AI glasses were released. The relevant person in charge of Xiaodu Technology has publicly stated that AI glasses are a very suitable interactive carrier for AI (artificial intelligence). Multi-modal signal input, data processing capabilities, intention understanding capabilities, and multi-modal interaction methods combining vision, voice and even eye movements and gestures have jointly driven the rapid development of the AI glasses industry.
Li Hongwei, founder and CEO of Thunderbird Innovation, publicly stated that Thunderbird’s layout logic is to use glasses as the core hardware entrance to provide AI interactive experience and prepare for the future AR era. Through our exclusive cooperation with Ali Tongyi, Thunderbird is continuing to invest in optimization of AI models, enhance product capabilities by developing AI applications, and form an ecological strategy of hardware +AI+ applications.
Third, in addition to helping the implementation of ecology and AI technology, another industry that AI glasses leverage is VR and AR companies when the metaverse exploded.
Some media have divided the current AI glasses into three categories: audio glasses that do not have display functions at all, AI glasses that integrate audio and camera, and AI glasses with AR functions. Representative companies of these three are Huawei, Baidu (Xiaodu), and AR companies such as Xingji Meizu, which became popular in the era of virtual reality explosion.
The advent of Vision Pro has allowed eye movement technology and AR display technology to accumulate experience in the hardware field. For glasses, the upgrade of display functions is of great significance to both users and large models.
Li Hongwei believes that the display function is highly intuitive, which is the key to human-computer interaction. For example, when people use AI glasses to navigate in a mall, if they see several restaurants, it is difficult to accurately distinguish which restaurant they point to based on voice commands alone. But if there is a display, the communication of information will become more intuitive and accurate.
2. Will AI glasses usher in a major explosion?
Compared with mobile phones and other smart wearable devices, the relationship between glasses and users is far closer than that between other devices.
Compared with smartphones and smart watches, glasses allow users to get rid of the trouble of looking down at the screen, and information can be presented in the user’s field of vision in real time, realizing the dream of raising their heads to socialize.
Moreover, the visual angle and visual range of human eyes far exceed the angle and range of traditional regular screens, which makes it easier to use glasses to assist people’s lives from different angles, and it is more convenient than cutting in from the screen.
Therefore, both companies and research institutions have increased the number of potential users of AI glasses to the level of the number of ordinary glasses users in the world.
A report released by XR industry research institute Weishen Information in 2024 shows that the current global sales of myopia glasses and sunglasses are approximately 1.56 billion pairs, and the global glasses market is approximately US$150 billion.It is predicted that 10 years later, global eye sales will reach about 2 billion pairs.Among them, the sales volume of myopia glasses + sunglasses in China is approximately 260 million pairs.In contrast, global sales of AI glasses are less than 2 million pairs.
In addition to the considerable user scale, the technical level also provides a lot of imagination for the future of AI glasses.
First, after superimposing AI, AR, and eye movement technologies, AI glasses have the possibility of replacing smartphones in theory.
The advent of Vision Pro has thrown a new way of interaction to the world. Users can interact with virtual reality through gestures, eye movements and voice commands, and the operation process requires no external controller.
It is worth noting that Cook originally hoped to directly launch a pair of glasses to replace the Vision Pro, but temporarily gave up because the technical conditions were not yet mature.
In September 2024, Meta launched Orion, its first AR glasses, inherited eye movement technology. Public information shows that it is a split design, adopts a waveguide design, has a thick black frame and weighs 98g. The glasses can be compatible with Meta’s self-developed EMG bracelet and have multiple interaction methods such as hand tracking, eye tracking, and head tracking.
The addition of AI allows the above devices to meet common functions of smartphones such as real-time translation, road navigation, calculators, and photography.
Before Orion came out, Zuckerberg said in a conversation with Huang Renxun that smart glasses would be similar to mobile phones and would be an always-on version of the next computing platform.
Li Hongwei also believes that just like smartphones, although there are still professional cameras, music players, smart watches and other devices on the market, mobile phones, as universal computing platforms, are fully compatible with these scenarios and create richer application possibilities. Further, AR glasses have the potential to become the next generation of general-purpose computing platforms.
Second, at the software level, the rise of DeepSeek has made AI glasses expected to reduce training costs with the help of low-cost and high-performance models, and then fly into the homes of ordinary people at a lower price.
In addition to DeepSeek, many manufacturers, including Byte’s bean bag model, have also updated their big models at the beginning of 2025.
In theory, the large model collects user data and needs through smart devices on the end side, learns and analyzes them, and then provides users with many help in daily life and realizes interaction with users. With the update of the large model and continuous progress in visual semantics and natural language applications, the efficiency of interaction is also constantly improving.
Northeast Securities Research News believes that DeepSeek’s open source features allow developers to make in-depth customization, and its API is low, which is conducive to the popularity of end-side AI.
In addition, Huatai Securities believes that the AI glasses platform, as a brand new consumer electronics category, logically does not exist like the smartphone ecosystem. Such brand manufacturers (Xiaomi, Apple) need to coordinate with existing Internet manufacturers (such as WeChat, Meta) to open up AI traffic. The situation of the portal makes it easier to become the first scenario for AI applications to be implemented.
3. The Triangle of Impossible is killing the first believers
In addition to embracing the opportunities brought by the development of AI, the foundation of AI glasses must be strong enough.
First of all, as a pair of glasses, weight, battery life and function are the primary criteria for users to choose or not.
Just like the experience that Vision Pro brought to users when it came out, its weight is close to 650 grams, which makes it quite difficult to wear. Relatively speaking, if an ordinary pair of glasses weighs 50 grams,, it is already a heavyweight product among ordinary glasses.
AI glasses manufacturers are working hard to bring AI glasses closer to the weight of ordinary glasses. For example, the AI glasses Meta Ray-Ban jointly launched by Meta and Ray-Ban weigh 49 grams. Thunderbird V3 parameters show that the glasses weigh 39 grams. As for the still rumored Xiaomi AI glasses, the currently exposed weight is 50 grams.
Behind the weight of just the same as ordinary glasses, battery life is also an important requirement of users. After all,If AI glasses want to win the market for ordinary glasses, AI glasses must be the same as ordinary glasses and can be worn by users for at least 10 hours a day.In this regard, whether it is the 4-hour battery life of Meta Ray-Ban and Rokid Glasses, or the nominal 7-hour battery life of the Thunderbird V3, there is still a gap between the use time of ordinary glasses.
As a result, users of Meta Ray-Ban often choose to buy two pairs of glasses at a time in case one pair runs out of battery. The Thunderbird V3 reserves a function to add a charging box to glasses, but while users wear glasses, they hang a glasses box underneath. This shape makes people laugh and cry.
Not only does the weight need to be lighter, the battery life needs to be longer, but there are also more functions needed.This is not a small challenge for hardware manufacturers. The screen, chip, and every function need to be increased in weight. There was once a view that there was an impossible triangle between function, battery life and weight.
Regardless of whether this impossible triangle can be overcome in the end, it at least reflects one reality. The prospects of AI glasses and the grand technical narrative are only one aspect. More importantly, the user experience level requires the joint efforts of all parties to solve a series of dirty tasks.
Faced with such dirty work, the next question arises, how to build a supply chain.
At present, the core competitiveness of major Internet companies is reflected in the big model, that is, the base level of the Internet ecosystem. The chips that determine computing power and the sensors, displays and other components that determine the experience need to be realized through cooperative OEM.
As for how to choose, each family has different practices. Based on comprehensive market information, Xiaomi AI glasses, which has yet to be officially released, have already announced cooperation with many suppliers, such as Qualcomm, Goer, Ofei Light, etc., which have been reported in the fields of chips, complete machine assembly, and cameras. Provide support for Xiaomi AI glasses. At the same time, Meta Ray-Ban was also completed by Goer.
This is related to Xiaomi’s current advantages. After Xiaomi determined a scientific and technological development strategy for people, vehicles and homes, the power of ecology has become Xiaomi’s core competitiveness. Squeezing into Xiaomi’s ecosystem is also a scene that various suppliers are happy to see.
Huawei is following the path of self-research. Although Huawei’s previously released smart glasses 2 only have audio functions such as AI intelligent broadcast for the time being, the outside world is optimistic about Huawei’s self-developed technical foundation in large models, chips, systems, etc., as well as its potential in subsequent market competition.
As for AR glasses companies, they have strengthened cooperation with traditional glasses manufacturers and large model companies. For example, the lenses of Thunderbird V3 were jointly developed by Thunderbird and lens company Dr. Glasses, and the AI technical support came from Ali’s Tongyi Thousand Questions Model. This is due to Thunderbird’s research and development strength and technical advantages in the core optical field of AR, and it also needs to leverage the advantages of other companies to empower AR glasses.
OEM or self-research is a question that needs to be carefully weighed for major Internet companies.The OEM model can allow large factories to go into battle with ease, but relying too much on external technology will inevitably fall into the dilemma of homogenization and insufficient innovation, and will not be able to form core competitiveness. The self-research model can have characteristics in subdivided fields, but the early research and development costs, especially before the impossible triangle is overcome, are always a lot of pressure.
The construction of the supply chain also determines the third key to the breakthrough of AI glasses, and how to coordinate with the existing ecosystem.
On the one hand, the matching of AI glasses with large models needs to be smoother. For example, in driving or walking scenarios, how to achieve rapid identification and rapid navigation at the millisecond level will not only affect the user experience, but also may affect traffic safety. Or, the integration of AI glasses and bone conduction headphones,
On the other hand, the configuration of AI glasses in terms of hardware and software also needs to be integrated. For major Internet companies that hope to use AI glasses to connect their products and form interconnection, the model and function of the chip are often not limited to AI glasses. For a product, it is necessary to comprehensively consider the parameters and configurations of all its equipment, which is a new challenge for the supply chain.
Northeast Securities Research News believes that looking forward to the future, the possible change trends for AI glasses are: 1) Using the addition of multiple batteries + single battery capacity expansion to improve battery life. The performance pressure and power consumption brought by end-side AI to hardware will increase consumer battery capacity. Capacity expansion has become an inevitable trend;2) Increasing the number of cameras to achieve better shooting results.
And this is only the first step in the comprehensive popularization of AI glasses for major Internet companies that have entered the game one after another.
In such exploration, the real norm for the growth of the AI glasses industry is the big waves sweeping away the sand and experiencing different stages of elimination and rebirth.
