CITIC Securities: Trends in smart glasses development from Ray-Ban Meta

According to the Zhitong Finance APP, in 2023, Ray-Ban will release the second smart glasses Ray-Ban Meta in collaboration with Meta. Compared to the first-generation product, the second-generation product will use the Qualcomm AR1 series chip and will be able to access Meta's Llama3 models to achieve AI voice interaction, AI object recognition, translation, and other functions. Zhongxin Securities released a research report looking ahead, suggesting a further iteration of glasses products at the September 2024 Meta conference; in the medium to long term, the trend of the AI phone industry has been established, and we are optimistic about the combination of AI and smart glasses/AR in the future, which may evolve into intelligent voice glasses, AI smart glasses, and AI+AR/MR glasses. It is recommended to focus on leading companies in the core components, modules, and complete machine links of the industry chain.

What are the highlights of the Ray-Ban Meta product?

As the second smart glasses co-produced by Ray-Ban and Meta, it is still positioned as smart voice glasses rather than AR glasses, with significant upgrades in interaction, chips, AI, and other functions compared to the first-generation product.

1. Interaction: Ray-Ban Meta has upgraded the camera resolution from 5MP to 12MP, and the microphone configuration from 3 to 5, enabling more accurate audio, image, and video information acquisition.

2. Chip: The first-generation product used the Snapdragon 4100+ main chip (a wearable platform released in 2020, 12nm process, integrated graphics processor), while the second-generation product uses the AR1 Gen1 main chip (an ARVR platform released in 2023, 4nm process, integrated NPU supporting edge AI processing); the storage chip has been upgraded from 4GB to 32GB capacity.

3. AI functions: Currently, it still needs to be connected to the Meta View App on the phone and only supports English conversations. Users can summon AI through the "Hey Meta" command, with AI functions divided into two categories. One category controls functions such as taking photos, making calls, and listening to music through AI, which can be achieved on the phone+glasses side (with relevant keywords embedded). The other category involves obtaining image and audio information through the glasses, transmitting it to the cloud for inference, such as identifying buildings, recommending recipes, real-time translation, etc., which requires the use of the cloud-based Llama3 model, resulting in some interaction delays.

4. Others: Ray-Ban Meta, as a smart glasses product, weighs only 49g, similar to regular glasses. Due to the lack of display function, there are various options available on the lenses such as myopia correction and blue light protection.

What is the pace and shipment volume of Ray-Ban smart glasses products?

In September 2021, Meta and Ray-Ban jointly released the first-generation smart glasses Ray-Ban Stories, with a starting price of $299. Due to limited usage scenarios (mainly for taking photos, videos, and listening to music), we estimate a total shipment volume of approximately 500,000 units throughout its lifecycle.

In September 2023, the second-generation product Ray-Ban Meta was released, supporting access to cloud-based large models. According to The Verge data, from September 2023 to May 2024, Ray-Ban Meta sold over 1 million units. We expect shipments to exceed 1.5 million units in 2024, and if promoted from the US and Canada to other regions worldwide, the shipments are expected to increase further We believe that the annualized shipment volume is expected to exceed 5 million units.

In addition, at the Q2 performance exchange meeting in early August, Meta also mentioned that the demand for Ray-Ban Meta currently exceeds production capacity. Meta will continue to increase production capacity to meet demand, and the company is also about to launch the next generation of Ray-Ban Meta products. Looking ahead, Meta is expected to continue advancing the iteration of eyewear products. We believe that products developed in collaboration with Ray-Ban will continue to focus on intelligent audio glasses with AI integration, extending from the original Ray-Ban products, while Meta products will focus on AR glasses with functions including optics and display.

What is the BOM cost and supply chain of Ray-Ban Meta smart glasses?

According to Wellsenn XR data, the BOM cost of Ray-Ban Meta is approximately $164, with the cost breakdown as follows: Qualcomm AR1 Gen1 costs about $55 (34%); ROM+RAM costs about $11 (7%); camera costs about $9 (5%); battery costs about $6.5 (4%); PCB costs about $7.2 (4%); and acoustic costs about $5.5 (3%). The production of smart audio glasses is relatively simple. According to the Meta performance exchange meeting, the manufacturer is EssilorLuxottica. In addition to smart glasses, we suggest paying attention to supply chain opportunities in future AR glasses. The added optical and display technologies in AR glasses have higher technical barriers, and BOM costs are expected to reach tens of dollars, with a significant increase in the overall value of the device.

How do we view the trend of smart glasses/AR glasses?

1. How do we view the smart glasses/AR industry? — With the establishment of the edge AI trend, we are more optimistic about the combination of AI and AR glasses after smartphones. Currently, the combination of AI and smartphones includes both edge models (smartphones with computing power) and cloud models. Consumers use smartphones more for voice and image interactions. We believe that devices worn on the head (such as glasses, headphones, etc.) will complement the shortcomings of smartphone interactions in the future. As AR glasses still face challenges in mass production and cost control, the form in which AI is first implemented may be smart glasses (combining vision and hearing) and headphones with cameras, with their computing power still relying on smartphones. Smart glasses, similar in form to regular glasses and weighing around 50g, can capture images/videos from the user's real perspective, achieving a "what you see is what you get" effect, enabling a more natural AI interaction.

2. How will eyewear products iterate? — Smart audio glasses → AI + smart audio glasses → AI + AR/MR glasses. ① Smart audio glasses can be understood as a combination of regular glasses and TWS earphones. The glasses connect to the phone via Bluetooth, allowing users to receive calls, audio, and other information from the phone, while also enabling voice control. Typical products include Huawei smart glasses. ② AI + smart audio glasses build on the first category of products, supporting AI integration for multimodal interactions, such as image recognition, real-time translation, etc., achieving more intelligent and continuous interactions. Typical products include Ray-Ban Meta. ③ AI + AR/MR glasses further upgrade the second category of products in display and optics, allowing for a combination of virtual and real information compared to smart audio glasses, which can only display real information We expect that Meta will launch similar products to define the product in the future.

3. How to view the shipment volume of smart glasses/AR glasses? As mentioned above, we believe that the computing power of smart glasses/AR glasses depends on smartphones or other terminals. However, the prices of AI smart glasses and AR glasses may be significantly different. The latter requires integration of optics, displays, and more difficult lenses, resulting in a higher price. Therefore, we believe that AI smart glasses and early-stage AR glasses are still positioned as accessories for smartphones or fashion items. Reference can be made to the shipment levels of TWS earphones (Counterpoint data shows that the shipment of TWS earphones in 2022 was 290 million pairs, accounting for approximately 24% of smartphones) and the shipment levels of glasses (Statista estimates that the global shipment of sunglass lenses will be approximately 850 million pieces in 2023, corresponding to about 425 million pairs; traditional lenses will be approximately 677 million pieces, corresponding to about 339 million pairs. A total of 760 million pairs). In the later stage, as the computing power of AR glasses improves and becomes independent of smartphones, becoming an independent computing platform, reference can be made to the shipment levels of smartphones (IDC data shows that the shipment of smartphones in 2022 was 1.2 billion units).

4. Which companies in the industry chain have more advantages? We are optimistic about top brands with hardware experience, a user base in mobile internet, and large model resources. In the long term, we believe that AI + smart glasses/ARMR is still a game for giants. ① As a mobile product, the user base is crucial. Smartphone brand manufacturers can expand based on existing smartphone users; internet companies can rely on existing mobile users. ② Smart glasses products as a new generation of hardware still need to cooperate with smartphones on the computing side. Therefore, companies with experience in smartphone production extending to glasses or companies with experience in mass production of smart glasses extending to the computing side of smartphones have more advantages. ③ Currently, at the hardware level of AR/MR products, improvements are needed in optics, optomechanics, heat dissipation, and battery life. At the application level, exploration is still needed in terms of content and landing scenarios. Companies with a foundation in content (social, office, video, etc.) and financial support for long-term research and development are more advantageous.

Risk Factors:

Lower-than-expected demand, slower-than-expected technological iterations, intensified market competition, weak macroeconomic environment, etc.

Investment Strategy:

We believe that glasses are another important intelligent carrier for AI, playing a crucial display function in interaction. The development path may present the process of intelligent voice glasses, AI smart glasses, and AI+AR/MR glasses. It is recommended to focus on leading companies in the industry chain that occupy core components, modules, and complete machine links