The entry of Xiaomi and Alibaba is changing the situation of the AI "hundred-mirror battle".

From an obscure geek toy initially to one of the emerging carriers for the implementation of large AI models today, the trend of AI glasses is gaining increasing momentum.

In the first half of this year, more than 10 types of AI glasses were officially released and exposed by the media. Among them, there are new products launched by startup players like Thunderbird, ThunderGod, and AnkerPrime, as well as new attempts made by established giants such as Huawei and Skyworth.

The shipment and sales volume of AI glasses have also witnessed a sharp increase.

According to relevant data from IDC, in the first quarter of 2025, the global shipment of smart glasses reached 1.487 million units, a year - on - year increase of 82.3%. It is predicted that the shipment of smart glasses in China is expected to reach 2.907 million units in 2025, a year - on - year increase of 121.1%.

JD.com data shows that during this year's 618 promotion, the sales volume of AI glasses increased by 7 times year - on - year, and the number of AR glasses users increased by 70% year - on - year.

The explosion of AI glasses is partly due to the reduction in weight. The new generation of AI glasses can weigh around 40g, similar to ordinary glasses. However, the main driving force is the large - scale application of large models.

Early AI glasses could only execute simple commands. The introduction of large models enables them to understand more complex contexts. Coupled with multi - modal fusion, they can perform more complex operations. For example, through "voice + vision + gesture coordination", users can "place an order with just a glance".

According to Lotu data, although in 2024, the sales volume of products with AI functions in the domestic smart glasses market accounted for less than 20%, it is expected that the penetration rate of AI functions in the smart glasses market will exceed 60% in 2025.

Among them, the influence of high - end players is becoming increasingly significant.

For example, Xiaomi and Alibaba, which recently launched AI glasses. The entry of these two giants is regarded as the turning point for AI glasses to move from niche toys to the mass market.

Xiaomi's Initial Exploration of the Ecosystem

In June this year, Xiaomi launched its first AI glasses at the "Xiaomi Smart Ecosystem Launch Event".

This AI glasses, which claims to be "comparable to Meta Ray - ban", has a battery life of 8.6 hours, exceeding the 4 - hour battery life of Ray - Ban Meta. Lei Jun defined it as "a personal intelligent device for the next era and a portable AI entrance". It can not only control Mi - Home devices at home, but also allow users to directly ask Xiaoai (Xiaomi's voice assistant) about anything they see, and Xiaoai will respond in real - time and provide answers.

However, the starting price of 1,999 yuan is only slightly lower than the $299 of Ray - Ban Meta. Although it has become the "price anchor" in the market, it is higher than the expected price of 1,499 yuan in the market.

Li Chuangqi, the person in charge of Xiaomi's wearable device department, later told the media that in order to make this project as successful as possible within a limited time, Xiaomi almost chose the most expensive components. Although there is a gross profit at the current price of 1,999 yuan, it is still a loss considering the overall investment.

The "limited time" mentioned by Li Chuangqi is about one year. Around the Tomb - Sweeping Festival in 2024, Lei Jun approved the official project establishment. And this opportunity was hard - won.

Xiaomi started to contact and research smart glasses as early as 2014, but it has always been a marginal business. Only about a dozen people have been in charge of this project for a long time. The three products that have been launched are all "functional glasses" rather than AI glasses in the current sense.

In 2023, Li Chuangqi first tried to apply to Lei Jun to develop AI glasses.

Li Chuangqi judged that the global glasses market itself is worth trillions of dollars, and the smart glasses market will be much larger. In addition, Xiaomi has supply - chain advantages in developing AI glasses. After the implementation of the "Smart Ecosystem Strategy", AI glasses can complement scenarios such as driving, home living, and health management, enabling seamless collaboration between devices. For Xiaomi, which is not good at content creation, adding AI provides an opportunity to become a new entrance.

Li Chuangqi's idea was a bit avant - garde for Xiaomi at that time. Xiaomi is good at developing product categories with available market data. In 2023, there was no consensus on AI glasses, and Ray - Ban Meta was not a hit.

More importantly, AI glasses are not isolated hardware but need to be integrated into Xiaomi's smart ecosystem. At that time, Li Chuangqi seemed to be lacking an opportunity.

At that time, Xiaomi's car SU7 was still in the critical testing phase of the engineering prototype, and it was still some time away from delivery. Without the ecological collaboration of the car, AI glasses could only serve as a mobile phone accessory and a home remote control.

This also explains why Lei Jun agreed to the project establishment when Li Chuangqi proposed it again around the Tomb - Sweeping Festival in 2024. By then, Xiaomi SU7 had been launched, and developing AI glasses could form the three core terminals of Xiaomi's ecosystem together with mobile phones and cars.

After the project was established, Xiaomi began to attach importance to this product. It not only used the "most expensive components" but also abandoned the original cooperation model with the ecological chain and instead let the group take charge directly.

This two - year - long perseverance received a positive response from the start.

On the night of the product launch, many Mi fans rushed to Xiaomi stores to pick up the limited number of in - stock products. The sales volume on JD.com exceeded 10,000 units within 12 hours of the launch, ranking first on the "Hot - Selling Smart Glasses List".

However, the return rate of Xiaomi's AI glasses began to rise after one month. According to the statistics of the industry media XR Vision, the return rate of Xiaomi's AI glasses may be as high as 40%.

Although Xiaomi used the best materials with great sincerity, it has to be admitted that Xiaomi's first AI glasses are rather mediocre.

The appearance is considered to have a "strong plastic feel and lack of fashion". The AI functions are also evaluated as a transplant of Xiaoai's existing capabilities, and many wake - up logics are almost exactly the same as those of Meta Ray - Ban and Huawei's smart glasses. The promotional slogan mentions the "Smart Ecosystem", but users actually experience a serious sense of fragmentation.

The most fatal issue may be the immature supply chain, which limits the gross - profit margin. The BOM (Bill of Materials) of the whole device is as high as 1,290 yuan, making it impossible to reproduce the "halved pricing" strategy used for mobile phones in the past.

This may be the reason why Lei Jun only spent 7 minutes introducing this strategic product, which is defined as an "AI Portable Entrance", at the product launch.

Compared with Xiaomi's previous practice of "setting benchmarks" for the first - generation mobile phones and cars, AI glasses seem to be a strategic move to occupy the position of the "ecosystem entrance" first.

Now it seems that Xiaomi's entry into the AI glasses market has a far - reaching impact on the industry beyond the product itself.

At least Lei Jun's participation has stimulated more technology and Internet giants to accelerate their self - research. The market has not cooled down but has instead witnessed more intense competition.

Alibaba's Ambition for Terminal Devices

One month after Xiaomi launched its AI glasses, Alibaba also announced its first self - developed AI glasses, Quark AI glasses.

Like Xiaomi, Alibaba also regards AI glasses as the most important personal mobile entrance after smartphones. The difference is that Xiaomi focuses more on hardware scenario - based applications, while Alibaba's AI glasses are an extension of software services.

In terms of functions, a pair of Quark AI glasses integrates almost the most core capabilities of Alibaba's ecosystem - the Tongyi Qianwen large model and Quark AI capabilities. It also supports functions such as Gaode Navigation, Alipay's "Scan - to - Pay", Taobao's "Shoot - to - Search", and Fliggy's travel itinerary reminders.

The media vividly described it as "wearing Alibaba's entire ecosystem on the face".

As a key move in Alibaba's AI - to - C strategy to extend from software to hardware, the emergence of Quark AI glasses represents the leap of Alibaba's AI capabilities from dispersion to concentration.

Previously, Alibaba's AI capabilities were scattered in individual independent apps, and users had to switch between different interfaces, making it difficult to form a unified intelligent experience. Based on the characteristic of being worn all day long, AI glasses can encapsulate Alibaba's capabilities in scenarios such as search, payment, and navigation into a single terminal, achieving "one - stop invocation".

For example, when users are traveling, the glasses can automatically synchronize Fliggy's travel itinerary, trigger Gaode Navigation, and recommend local - life services. This cross - business active invocation breaks the traditional app - isolation model.

As is well - known, Alibaba does not have a deep - rooted hardware - product gene. Using AI glasses as a physical terminal not only integrates software capabilities and improves the user experience but also completes Alibaba's complete chain from large AI models to applications and then to hardware.

Hardware manufacturing is not an overnight thing. Facing the shortcomings and hardware challenges in the supply chain, Alibaba's solution is to use a definite scale to drive suppliers in the reverse direction.

On the one hand, Alibaba forms clear user needs and usage scenarios through high - frequency and essential scenarios within the ecosystem, guiding the supply chain to adapt with the value of the ecological closed - loop, rather than relying solely on the iteration of the hardware itself. On the other hand, it guides the maturity of the industrial chain with demand and orders.

Song Gang, the chief architect of Quark AI glasses and the person in charge of the intelligent terminal business of Alibaba's Smart Information Business Group, also said in a media interview that "when there is a large enough order volume and clear technological requirements and trends in this industry, suppliers will be motivated to make technological breakthroughs and invest in production capacity. In the future, a separate glasses supply chain may be formed."

Currently, the supply chain of AI glasses has a high degree of overlap with the mobile - phone supply chain, and Alibaba is promoting the formation of an independent glasses supply chain "such as components like optical engines, optical waveguides, and wearable chips". In addition, when integrating the design and supply chain with traditional glasses brands, as all manufacturers are on the same starting line, Alibaba can exert its strength in this field.

Since the end of last year, Alibaba has carried out a series of internal adjustments: the AI application "Tongyi", originally belonging to Alibaba Cloud, was split and incorporated into the Smart Information Business Group where Quark is located. At the same time, it promoted the integration of the Tmall Genie team and the Quark product team to explore new hardware directions such as AI glasses. Wu Jia, the CEO of the Smart Information Business Group, coordinates Alibaba's AI - to - C applications.

According to the plan, the standard version of Quark AI glasses is priced at 2,999 yuan, basically equal to the cost. In Wu Jia's words, "We don't make money by selling glasses."

However, it is speculated that Quark's real profit point may lie in the expansion and use of more scenarios for the glasses. For example, in the e - commerce field, for each successful "price - comparison" transaction made through the glasses, Alibaba can charge a certain technical service fee from merchants.

In addition, the "Enterprise Edition" for the B - end has also started internal testing. DingTalk will package functions such as meeting minutes and remote expert guidance and charge based on the annual seat - fee system.

From the current situation, integrating Alibaba's ecosystem is Quark's core advantage, but it may also be a double - edged sword: Will the deep - seated binding with Alibaba - affiliated apps limit the non - Alibaba user group?

Of course, more importantly, compared with Xiaomi, Alibaba has significantly less experience in the hardware industry. This is a great test for Wu Jia.

The Intensifying Competition in the AI - Glasses Market

In 2025, the competition in the AI - glasses market is intensifying, and the number of players is still increasing.

In addition to Xiaomi's supply - chain advantages and Alibaba's huge ecosystem, each player entering this field has its own "unique skills".

For example, Huawei has achieved low - latency collaboration among multiple devices such as mobile phones, screens, and AR glasses through "NearLink + HarmonyOS", supporting the simultaneous connection of more than 10 devices with a latency as low as the microsecond level. Another example is Meta, which firmly holds the top position overseas with its global content library and CV algorithms.

Among startup players, Thunderbird has achieved a mass - production yield of 92% for its self - developed MicroLED optical engine, and the yield of the new - generation optical engine has been further increased to 95%. XREAL's 57 - degree wide - field - of - view and 200 - inch virtual giant screen lead in the viewing experience.

However, behind the bustling scene of the hundred - player competition, there are still many real challenges that are difficult to overcome.

An urgent problem is the chip. For AI glasses to have a fast feedback speed and clear photos, they need a powerful main - control chip. However, most of the chips currently used in AI glasses have high power consumption and high costs, and their performance, cost, and power - consumption indicators are not ideal.

Some main - control chips even account for 31% of the cost of smart glasses. Taking Qualcomm's AR1 as an example, the cost of a single chip is about $55, directly limiting the product - pricing space.

This is why although Xiaomi's AI glasses at a price of 1,999 yuan have a gross profit, it is still difficult to cover the R & D and supply - chain investment. This is even more so for the entire industry, which generally faces the paradox of "the more you sell, the more you lose".

In addition, the lack of computing power at the edge side further affects the user experience. Edge - side AI has difficulty balancing high computing power and low power consumption. AI glasses currently rely on cloud - based AI models and computing power, resulting in a slow feedback speed. If used continuously for a period of time, they will consume a large amount of power and generate heat.

Coupled with the insurmountable "impossible triangle" - performance, light weight, and long battery life.

Undoubtedly, AI glasses cannot be developed with only a single - point advantage but require a long - term systematic project. The advantages of large enterprises seem to be more obvious.

Li Chuangqi previously mentioned the view that "only two types of manufacturers can succeed in AI glasses: mobile - phone manufacturers and national - level app - application manufacturers. Large enterprises have traffic, content, and large models, and all they need is a hardware carrier."

With strong financial support, Xiaomi has adopted a dual - chip solution of Qualcomm AR1 + Hengxuan 2700. The Qualcomm AR1 is used in complex scenarios, while the Hengxuan 2700 chip is used in simple scenarios such as listening to music. Currently, many domestic AI - glasses manufacturers still use a single - chip solution.

Large enterprises also have advantages in brand trust and sales channels.

Glasses are different from other smart products and have a certain dependence on offline channels. Rokid previously claimed that it will cover more than 1,000 offline glasses stores by the end of this year, and its shipment target for next year is 1 million units.

For example, Xiaomi's offline "Xiaomi Stores" exceeded 10,000 at the end of 2023. Coupled with its online self - operated e - commerce platform, customers can experience the products before making a purchase.

The competition in software is even more brutal. The level of AI - glasses software capabilities mainly depends on whether it can smoothly invoke various apps in mobile phones. This depends on