Huachuang Securities: AI glasses are rapidly iterating, and the industry may be on the verge of breakthrough.

Hua Chuang Securities released a research report stating that the popularity of Ray-Ban Meta smart glasses has opened the prelude to the rise of the AI glasses industry. Currently, various brands are entering the industry, which has entered a stage of vigorous development. Hardware upgrades and iterative improvements in large models may allow glasses to replicate the "smartphone moment"; the good results of Deepseek-R1 distillation and open source provide good conditions for edge deployment, and may help solve pain points in the AI glasses content industry. Continuous product innovation may help drive rapid growth in terminal shipments, and the industry may have entered the first year of explosion, with related targets in the industry chain expected to benefit fully. The main points of Hua Chuang Securities are as follows: Ray-Ban Meta smart glasses have opened the rise of the AI glasses industry, and various brands entering may drive 2025 to become the first year of explosive growth in the AI glasses industry In September 2023, Meta launched the new generation Ray-Ban Meta smart glasses, which attracted wide attention from its stylish appearance and linkage with platforms like Instagram. Sales have been climbing quarterly after integrating multimodal large models in Q2 2024, with single-quarter sales reaching 680,000 pairs in Q4 2024. Up to Q4 2024, Ray-Ban Meta has accumulated sales of 1.63 million units, and the popularity of Ray-Ban Meta has also opened the rise of the AI glasses industry. Since December 2024, many brands including Dapeng, Leibird, Rokid, Liweike Technology, etc., have held AI glasses launch events and released their own new AI glasses products. Additionally, leading consumer electronics manufacturers such as Xiaomi, Samsung, Meta, and Apple are also preparing related products. Continuous product innovation may drive the rapid maturation of AI glasses and promote growth in terminal shipments. According to Wellsenn XR data, global AI glasses sales were 1.52 million pairs in 2024 and are expected to grow to 3.5 million pairs in 2025, a 130% increase. The rapid innovation and iteration of the industry + rapid growth in terminal sales may make 2025 the first year of explosion for AI glasses. Hardware upgrades and iterative improvements in large models may allow glasses to replicate the "smartphone moment" (1) In terms of hardware, consumer electronics innovation is a driving force, with AI glasses adding components such as cameras, microphones, storage, SoCs, etc., compared to traditional glasses, enabling functions such as voice interaction, photography, etc. Additionally, by integrating large models, users can further enhance their experience by controlling various functions through voice commands. (2) In terms of model capabilities, since 2023, generative artificial intelligence applications such as Siasun Robot & Automation Chat GPT, AI programming tool GitHub CoPilot, and image generation system Stable Diffusion have emerged, bringing new experiences in text creation, image generation, code generation, etc. In January 2025, DeepSeek-R1 was released and simultaneously open-sourced. Its performance in tasks such as mathematics, code, and natural language reasoning was comparable to the official version of OpenAI o1. Additionally, the distilled output of the DeepSeek-R1 model achieved performance similar to that of OpenAI o1-mini in multiple capabilities. The effectiveness of AI models directly determines the user experience of AI glasses. The good results of Deepseek-R1 after distillation and open-source provide good conditions for edge deployment and may help solve pain points in the content end industry of AI glasses. AI glasses have multiple upgrades, and incremental links such as SoCs, SiPs, cameras, and batteries may benefit Compared to the cost structure of traditional glasses, SoCs, cameras, batteries, and other components are the main increments for AI glasses, with cost proportions of 33.54%, 5.49%, and 3.96% respectively (using Ray-Ban Meta smart glasses as an example). SoCs are the core brain of AI glasses; photography and video recording are one of the main functions of AI glasses, with current AI glasses cameras mainly focused on single cameras with 12-16 million pixels. The quality of images determines the accuracy of AI functions, so cameras may be further upgraded. AI glasses face a bottleneck in battery life, but solid-state batteries and silicon-carbon anode batteries can significantly improve battery energy density and potentially increase the battery life of AI glasses once the technology matures. In addition, the weight and size of glasses are major concerns for consumers, and the trend towards lightweight and compact AI glasses has become an industry consensus. SiPs can integrate different components to reduce the weight and size of glasses, making it a solution for future AI glasses. Although AI glasses currently do not primarily include display solutions, image display can greatly improve information acquisition efficiency, and AI+AR remains the core development direction for smart glasses. Risk warning: Product penetration and consumer electronics demand may not meet expectations.