How far can AI glasses go?

Abstract

The product has undergone multiple rounds of validation, and AI glasses have shown successful paradigms. 1) Pricing: The price is affordable, and there is still room for cost reduction in the future. Global AI glasses brands have successively released multiple products, with prices generally in the range of 1,000 to 3,000 yuan, providing users with a lower trial cost. In addition, we are optimistic about future cost reduction opportunities for core components such as SoC, opening up lower-tier markets and attracting users to purchase; 2) Interaction and Applications: We believe that smart glasses connected to large models provide a new interaction method, while large models offer more application scenarios, promising to create an all-scenario life assistant and provide a new hardware experience. 3) Brands: Currently, AI glasses products have been released by Meta, Baidu, Rokid, and Xreal, and we expect that major companies such as Samsung, Apple, ByteDance, and Xiaomi have related product plans, with products likely to be launched in the next 1-2 years. We believe that new terminals require global leading companies to educate consumers, complete market promotion, and attract more application developers to enrich the ecosystem. Looking back at the multiple rounds of industrial development in ARVR, we believe that the successful paradigm of AI glasses has emerged, and the industry is expected to enter a rapid development phase.

AI glasses still have significant optimization space, and the rich product forms are expected to drive shipment growth. 1) Basic Functions: We believe that future AI glasses are expected to solve users' customized refractive solutions through ultra-thin laminated lens designs, and improve the adaptability of wearing sunglasses through electrochromic technology, helping AI glasses achieve and optimize basic functions. 2) SoC: According to the current product BOM disassembly, SoC occupies a high cost, and we are optimistic that domestic chips are expected to be released successively, providing more cost-effective chip solutions and upgrades in terms of battery life and cost. 3) Storage: We are optimistic that future smart glasses will have rich functions, which are expected to drive the upgrade of storage specifications, and with the growth of shipment volume, we estimate that the future storage market is expected to achieve a market increment of billions of dollars. 4) Optical Display: In the medium to long term, we believe that optical display is one of the key upgrades for AI glasses. As core components such as waveguides and Micro LED mature, global leading brands are expected to be the first to launch AI glasses (AR glasses) equipped with displays, further enriching application scenarios and completing information interaction and feedback through display functions, optimizing the user experience. In summary, we believe that in the next 3-5 years, AI glasses are expected to achieve continuous upgrades in basic functions, SoC, storage, and display, improving user experience, enriching application scenarios, and driving overall product shipment growth.

Main Text

AI+ Glasses are Expected to Create the Next Mobile Terminal

AI glasses may become the optimal solution for multimodal large models

AI glasses are compatible with multimodal large models and are expected to lead industry innovation trends. With the release of ChatGPT-4o and Doubao visual models, multimodal AI is accelerating its implementation, and interaction methods such as voice, video, and images are expected to refresh the market's perception of AI capabilities We believe that AI/AR glasses products possess 透视能力 (透视能力), and the devices are equipped with various sensors such as cameras, microphones, and speakers, enabling real-time material acquisition and interaction through voice or display, which is expected to become a better choice for multimodal AI terminals.

In 2024, AI glasses brands will flourish, and more major manufacturers are expected to enter the market in 2025. This will achieve multimodal AI upgrades, with multiple global consumer electronics brands releasing related products, continuously increasing the attention on the AI glasses market. Looking ahead, we expect that in 2025-26, more leading consumer electronics brands will likely launch related products, continuously investing in AI glasses terminals, which are expected to create the next generation of AI smart wearables.

Chart 1: Review and Outlook of Consumer Electronics Terminal Upgrade Trends

Chart 2: Summary of AI/AR Glasses Releases from 2024 to 2026 (Partial)

There is significant penetration potential, and the future growth of AI glasses is promising.

By 2030, shipments of AI smart glasses are expected to reach 80 million units. We believe that as the functions of smart glasses become richer, future features such as refractive adjustment and color-changing are expected to gradually replace the traditional myopia glasses and sunglasses market. Additionally, with the decline in raw material costs, the penetration rate is expected to accelerate. According to Wellsenn's calculations, global shipments of AI smart glasses are expected to reach 80 million units by 2030, with a penetration rate of approximately 4.3%.

Chart 3: Shipment Volume and Penetration Rate of AI Glasses

Why AI glasses?

Product forms have undergone multiple rounds of validation and feedback.

ARVR has been refined through various product forms, and smart glasses may be the optimal solution. Since 2010, market attention to ARVR has continuously increased, with major tech companies such as Apple, Microsoft, Meta, Samsung, Google, and ByteDance entering the field, as well as the rise of startups like Magic Leap, Xreal, and Rokid. We believe that the product forms have undergone multiple iterations and functional innovations, and the successful paradigm of ARVR product forms is now relatively clear. We are optimistic that smart glasses are likely to accelerate the development of the ARVR industry.

Chart 4: ARVR has undergone multiple product validations and customer feedback

AI glasses are lightweight and easy to use, creating an Always-On smart terminal experience

Simple functions with low user learning costs. Taking Meta-Rayban glasses as an example, AI glasses are based on traditional glasses, adding features such as cameras, acoustics, and AI large model access, becoming AI smart glasses. The main functions of the product focus on video and photo shooting and voice interaction. We believe the functions are relatively simple and easy for users to operate, leveraging traditional terminals like smartphones as a support, and through Bluetooth connectivity, can place subsequent complex operations and materials on traditional terminals.

Large model empowers new terminals, with comprehensive application scenarios. We believe that integrating large models enriches the functions of AI glasses, achieving full coverage of application scenarios. In the future, as domestic large models mature, it is expected to accelerate the growth of terminal shipments.

Chart 5: AI glasses have simple functions and rich application scenarios

There is significant room for cost reduction, and products are expected to continue to penetrate the market

The cost structure of smart glasses is similar to that of TWS, with cost reductions driving product penetration. Looking back at the growth history of TWS, we believe that after Apple AirPods set the industry standard, the rapid cost reduction of core components has been one of the key factors driving product shipment growth. Similar to TWS, the main components with high cost proportions in smart glasses are SoC, storage, and PCB-related products. We are optimistic that more domestic suppliers will enter the market for the aforementioned components, replacing current high-end chips like Qualcomm AR1 with customized, low-cost products, which is expected to drive down the cost of smart glasses and lead to a decrease in product prices, potentially driving continuous growth in shipment volumes.

Chart 6: ARVR BOM Comparison (2024)

How to view the future development trend of AI glasses products?

Lightweight AI glasses are becoming a development trend, and future overlay display functions are expected to become independent terminals. Currently, AI glasses only have voice interaction and shooting functions, but they are lightweight and affordably priced, leading to strong user purchase intentions. The stylish product design also serves a decorative function, making them the mainstream product in AR glasses today. Looking ahead, we believe that in the next 3-5 years, lightweight AI glasses will continue to be the main trend in industry development. As the costs of core optical components decrease, we believe that display functions are expected to gradually enrich, achieving a more comprehensive independent AI interaction terminal Chart 7: Outlook on the Development Trends of AI/AR Glasses

Prescription lenses address user pain points, electrochromic technology provides sunglasses solutions

Smart glasses may primarily target traditional glasses users, with clear demand for refractive correction. We believe that the potential users of smart glasses are still mainly traditional refractive correction or sunglasses users. Looking ahead, we think the supply chain for prescription lenses in smart glasses may participate in the industry chain supply, addressing users' refractive adjustment needs through ultra-thin lenses, which will then be sent downstream for assembly, ultimately delivered to consumers as finished products.

Electrochromic technology provides sunglasses solutions for smart glasses, with mature related technology layout. We believe that sunglasses are an important function aside from vision correction glasses. Traditional glasses often use specific colored lenses combined with coating processes to achieve sunglasses effects. As electrochromic technology matures, products like Xreal Air have already incorporated related functions. We are optimistic that electrochromic technology is expected to penetrate smart glasses in the future, thus addressing users' sunglasses needs, attracting purchases, and promoting the penetration rate of smart glasses.

SoC cost accounts for a high proportion, with a clear upgrade trend

The value of the main control chip accounts for over 60% of the mainboard cost and is a major upgrade direction. According to the teardown report from Wellsenn XR, the electronic materials of the Ray-Ban Meta glasses mainboard cost about $88.5, with the Qualcomm main control AR1+ chip set (including WiFi and power management chips) costing around $58, accounting for 62%. Additionally, from the perspective of upgrade trends, we believe that performance improvement and power consumption optimization of SoC are also the main chip technology upgrade directions, aimed at solving the current bottlenecks of simple product functions and insufficient battery life. We believe that the value proportion of SoC in the BOM is expected to continue to increase in the future.

AI glasses SoC can be simply understood as smart headphone chip + ISP chip. As a system-level chip, SoC consists of numerous functional modules, with different scenarios presenting differentiated demands for SoC chips. As a wearable device, AI glasses are limited by edge computing power and power consumption, and we believe that they will not be able to run local models for a considerable time in the future. Furthermore, issues such as unstable network connections and long battery life currently significantly affect user experience. Therefore, we believe that the AI computing power demand for the main control chip of AI glasses is relatively low, focusing more on connectivity and low power consumption capabilities. Additionally, since AI glasses have added cameras compared to traditional smart audio glasses, and with the development and maturity of video understanding models, we believe that multimodal interaction will become standard for AI glasses, making image processing (ISP) capability crucial.

We are optimistic about domestic SoC companies gradually completing domestic substitution within the next two years, mainly for the following reasons: (1) From the perspective of underlying technical capabilities, domestic SoCs each have their advantages, with cost-effectiveness and power consumption becoming the biggest breakthroughs

As wearable products, lightweight AI glasses do not require the extreme functionalities of smartphones, but rather need chips that are balanced in performance, battery life, and cost-effectiveness. We see that the combination of smart headphone chips + ISP chips may become the mainstream of the first generation of domestic solutions. Even with the use of two chips, we believe that domestic solutions have sufficient competitiveness in terms of cost-effectiveness. We are optimistic that the next generation of domestic manufacturers' products can also provide highly integrated chip solutions.

Chart 8: Layout of AI glasses chip suppliers

(2) From the perspective of product definition, domestic leading chips are more customizable and have strong scene reusability

Currently, we have compared three chips: Qualcomm AR1, UNISOC W517, and Hengxuan BES2800. As shown in the table below, although Qualcomm AR1 has powerful performance, it seems somewhat redundant for lightweight AI glasses applications. For example, Qualcomm AR1 provides a display capability of 1.6 million pixels per eye, while most AI glasses do not require display functions. The use of WIFI7 has not yet been popularized in the mobile phone and router fields and currently does not have the conditions for widespread use. UNISOC W517 is an old product from four years ago, primarily used in fields like smartwatches, while Hengxuan is more proactive in its layout for smart glasses.

Chart 9: Comparison of mainstream chip solutions for AI glasses

(3) From the downstream perspective, domestic companies have advantages with large customers in the internet and Android smartphone brands

Large models empower the hardware ecosystem at the edge, and domestic internet companies are more proactive than overseas internet giants in laying out the hardware ecosystem. Upstream, we have increasingly mature domestic multimodal large models; midstream, we have various SoC chips that meet various complexities and functional requirements, along with a strong module and assembly supply chain; and downstream, we have the advantage of terminal brand customers that have long been in close cooperation. We believe that when innovative terminals emerge, domestic leading SoC companies also have the first-mover advantage of entering pre-research with terminal customers early.

Rich functions of glasses enhance storage specifications, and shipment growth brings incremental space

Storage chips are the semiconductor devices with value second only to the main control chip in AI glasses. From the two "hot-selling" products that have already been launched, Ray-Ban Meta glasses and Shanjing AI camera glasses, both use Baiwei Storage's ePOP chip. ePOP (embedded Package on Package) is a storage product that combines NAND Flash and LPDDR, widely used in smart terminals with strict requirements for chip size, especially in the field of smart wearable devices Chart 10: AI Glasses Disassembly & ePoP Packaging

Currently, AI glasses primarily use a combination of 2GB DRAM + 32GB NAND Flash for storage. We believe that as the functions of AI glasses become more diverse in the future, the demand for storage will also increase. Taking Ray-Ban glasses as an example, the first-generation product released in 2021 used a storage configuration of 512MB LPDDR3 + 4GB eMMC, while the second-generation product in 2023 has upgraded to 2GB LPDDR4X + 32GB eMMC, showing a significant upward trend. We believe that with the improvement in performance of AI glasses in areas such as imaging, audio, and local AI, the storage value per unit will further increase in the future.

Chart 11: Evolution of Ray-Ban Glasses Storage Solutions

We estimate that the current demand for storage in AI glasses may be in the tens of millions of dollars, and in the coming years, as sales of AI glasses exceed ten million units, the market size is expected to reach the hundred million dollar level. In the medium to long term, if sales of AI glasses exceed one hundred million units, we believe that their impact on the storage market could drive orders to tens of billions of dollars or even exceed one hundred billion dollars. We believe that with advantages in supply chain and cost-effectiveness, companies in mainland China are expected to gradually increase their market share.

Chart 12: Estimation of AI Glasses' Impact on the Global Storage Market

Smart glasses are expected to increase display solutions in the future, focusing on optical innovation.

Waveguide lenses are the future development trend for AI glasses. Currently, the more mature optical imaging solutions in the market include prisms, off-axis reflection, freeform surfaces, Birdbath, and waveguide solutions. Looking ahead, we believe that waveguides can resolve the conflict between field of view and product size, and while reducing the size and weight of the device, they are more similar in form to traditional glasses design, aligning with the characteristics of AI glasses. Therefore, we are optimistic about the development potential of waveguides, which are expected to achieve mass production in collaboration with industry giants and startups, and help AI glasses products add display functions.

Chart 13: Comparison of AR Optical Display Solutions

Chart 14: Comparison of Optical Waveguide Process Paths

Micro LED has significant volume advantages and is expected to become a display solution for AI glasses when paired with optical waveguides. The current AR light source solutions proposed in the market mainly include LCOS (Liquid Crystal on Silicon), DLP (Digital Light Processing), LBS (Laser Beam Scanning), Micro OLED, and Micro LED solutions. We believe that lightweight is one of the important advantages of AI glasses. Micro LED has significant advantages in terms of volume, power consumption, brightness, and lifespan, making it more suitable for AI glasses. We are optimistic that in the future, with a decrease in costs, it is expected to be paired with optical waveguides to become a display solution for AI glasses.

Chart 15: Comparison of AR Optical Imaging Solutions

Authors of this article: Li Chengning S0080522050003, Zang Ruocheng S0080522070018, etc. Source: CICC Insights, Original title: "CICC | ARVR Series #7: How Far Can AI Glasses Go?"

Risk Warning and Disclaimer

The market has risks, and investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk