AI/AR Glasses Industry Research Report, 2025

ResearchInChina released the "" AI/AR Glasses Industry Research Report, 2025"", which deeply explores the field of AI smart glasses, sorts out product R&D and ecological layout of leading domestic and foreign manufacturers, and analyzes OEM-related eco-chain enterprises in AI glasses field. It conducts a panoramic review of typical AI glasses products newly launched between 2024 and 2025, and presents their technical highlights and functional characteristics in multiple dimensions.

1. In the past 12 months, mainstream manufacturers have released more than 30 AI smart glasses

Between 2024 and 2025, many manufacturers at home and abroad have launched AI smart glasses, especially around the CES2025, the market ushered in an intensive release period. Market statistics show that in 2024, the global sales of AI smart glasses were about 1.52 million pairs, of which AI glasses brands in China accounted for about 3%, or only 50,000 pairs. As major manufacturers continue to exert their efforts in technology R&D and market promotion, the AI smart glasses market has great potential for future growth. It is estimated that by 2030, the global sales of AI smart glasses are expected to soar to 90 million pairs, becoming a growth point that cannot be ignored in the field of consumer electronics.

From a technical perspective, some manufacturers have incorporated AR display functions into AI smart glasses, cleverly combining virtual information with real scenes and presenting them to users in a more intuitive and vivid way, effectively broadening the application scenarios and functional scope of AI glasses.
In terms of weight, the overall weight of AI glasses varies greatly, ranging from as light as 25g to 99g. The weight of most products is concentrated between 31g and 70g. The trend of lightweight reflects the significant improvement in hardware integration.

2. Five technical paths for AI glasses foundation models, with more than 30 mainstream models

AI glasses are increasingly connected to foundation models. From the cooperation between AI glasses manufacturers and foundation model suppliers, there are mainly three core strategies:

① Completely independent research and development of foundation models: This type of glasses manufacturer needs to have the ability to independently build foundation models, aiming to achieve a closed loop of technology and ensure the autonomy and controllability of the technology.

② Access to third-party services: Through the ability to call external foundation models through API or SDK, glasses manufacturers can quickly integrate functions. Some manufacturers choose to deeply customize their own foundation models to optimize the interactive experience; while others build a multi-model ecosystem to give users the flexibility to call on demand.

③ ""Self- developed + access"" hybrid modes: This strategy aims to balance the autonomy of technology and commercialization efficiency. In key scenarios, manufacturers will independently develop models to meet specific needs; for general functions, they will access third-party foundation models to achieve rapid deployment.

In short, AI glasses use a variety of methods when equipped with foundation models, including deep integration of self- developed foundation models, customized models for specific scenarios, compatible calls of multiple models, end-cloud collaborative strategies to optimize performance and power consumption, and lightweight technology to achieve local deployment.

3. AI smart glasses ecosystem

AI glasses are based on the ""all-in-one integration"" mode. By integrating core components such as speakers, cameras, biosensors, AR displays, etc., they cleverly combine functions of multiple devices such as myopia glasses, smart speakers, sports cameras, e-books, projectors, etc., to meet users' needs in audio-visual, photo, health monitoring, information interaction, etc., bringing users a convenient, diverse and efficient experience.

AI glasses are gradually building a comprehensive, intelligent ecological interconnected environment for users. Its core value lies in being user-centric, relying on efficient interconnection technology and hybrid computing architecture to seamlessly connect various terminals in life, work, and entertainment (such as smartphones, computers/keyboards, game controllers, smart home devices, and IVI terminals).

Kanaan Technology officially released the KANAAN-K1 AI smart glasses on December 31, 2024. The glasses are equipped with voice assistant ""Xiaonan"", which can easily connect to various terminals. Users only need to wake up ""Xiaonan"" to seamlessly access smart home ecosystem, freely adjust the brightness of indoor lights, the temperature of air conditioners, and the opening and closing of curtains, and fully realize the convenient experience of ""voice as a service"".

4. Application prospects of AI glasses in vehicles

DreamSmart's AR+AI smart glasses, powered by the Flyme AI OS and Flyme Link technology, enable seamless connectivity and smooth switching across multiple devices (including smartphones, IVI systems, and other terminals) in all scenarios, delivering a boundless intelligent experience.

AI glasses technology is accelerating its iteration, and the standardized system platforms and interconnection technologies built by major manufacturers are becoming more mature, which gives AI glasses a broad application prospect in automotive field.

As seen from the above functions, AI glasses and AR-HUD share some overlapping functionalities. However, AI glasses far surpass AR-HUD in terms of virtual display size, display richness, multimodal interaction capabilities, and upgradability.

Beyond enabling ""display enhancement"" in the cockpit, AI glasses can also provide ""computing power enhancement."" With the advancement of foundation models, smart cockpits are demanding increasingly higher computing power. Upgrading the built-in IVI system's computing capacity is often challenging, whereas some AI glasses come equipped with a computing power module that can enhance the cockpit's processing capabilities.