From Google Glass to Vision Pro, the exploration of smart glasses has seen numerous attempts yet to achieve widespread success. It wasn't until Ray-Ban Meta emerged that newfound hope resurfaced in this industry. As we enter the second half of 2024, the deepening application of large AI model technology has breathed new life into the realm of smart glasses.

Collaborating with Rokid, a leading domestic smart glasses brand, Baidu has launched AI-enabled smart glasses. Additionally, rumors suggest that Xiaomi and Apple are joining the ranks of smart glasses development. Numerous technology giants, including OPPO, vivo, Huawei, Tencent, and ByteDance, are actively evaluating AI glasses projects, hinting at a burgeoning smart glasses market.

Currently, the sales volume of AI glasses is minuscule compared to the colossal scale of the smartphone industry. So, what is the crucial catalyst to further propel the penetration rate of AI glasses?

01. AI Glasses: Still a Work in Progress

Over the past year, news of over a million sales of Meta Ray-Ban glasses has excited many companies. The promising new category of 'AI glasses' has once again become a hot topic of discussion. In the streets of the United States, it's common to see people wearing these dual-camera and audio-equipped glasses, showcasing the fusion of technology and fashion.

So-called 'AI glasses' are essentially glasses equipped with audio and camera functions. This concept is not novel; years ago, Bose's BoseAR glasses already featured audio functionality, akin to sunglasses-shaped headphones. Snap also launched the first generation of Spectacle glasses with cameras, enabling users to effortlessly shoot short videos. However, why have 'AI glasses' suddenly regained popularity?

The reason is straightforward: the advent of large models.

Currently, the primary concept is to utilize large models on mobile phones (typically through an app like ByteDance's Doubao) to 'empower' glasses, transforming them from simple wearable devices into true AI smart hardware.

However, in my view, the current 'AI glasses' will inevitably transition to AR glasses in the short term, evolving from 'audio + camera' to 'audio + camera + near-eye display.' In practical use, when forced to reach for your phone, 'AI glasses' lose their significance as an independent smart device.

For a product, the context is crucial. Take Baidu's Xiaodu AI glasses as an example; they are touted to have functions such as first-person-view recording, asking questions while walking, calorie recognition, object identification encyclopedia, audio-visual translation, and smart reminders. These functions sound enticing, but upon closer examination, they can all be performed on a mobile phone. AI glasses essentially shift the audio and camera from the phone to your head, facilitating use in specific scenarios (like navigation and photography when your hands are occupied). However, this doesn't fully demonstrate the unique advantages of AI glasses.

In practical use, most interactions between users and AI still occur on mobile phones. When you need information, you ask AI, which sends a lengthy text response. However, on AI glasses, this interaction becomes cumbersome. Imagine wearing AI glasses and asking questions while walking; you might not remember or have the patience to listen to AI's answers. Since the brain processes text or image information much more efficiently than sound, you prefer to scan content with your eyes and swiftly move on to the next question. Additionally, voice interaction suffers from latency; you must complete the entire sentence before AI can comprehend and respond, often leading to incorrect or irrelevant answers.

Regarding the camera function, while the camera on the glasses offers an elegant photography option, there are many inconveniences in practical operation. When taking photos, you still need to refer to the frame in the phone's viewfinder, and often, you may need to squat down or adopt various awkward angles to capture a good shot. This is nearly impossible with current AI glasses, not to mention that their camera performance is inevitably much inferior to that of mobile phones.

Therefore, I believe that for AI glasses to truly become popular and evolve, they must transition to AR glasses with the addition of near-eye display functionality. Only then can AI glasses genuinely leverage their unique advantages and emerge as an independent smart device from mobile phones, bringing more convenience and fun to people's lives.

02. The Transition to AR: A Potential Catalyst

The development of smart glasses has gradually differentiated into three distinct technology pathways: AR-functional smart glasses, AI glasses without a display screen, and audio glasses focusing on audio experience. Currently, the core difference between AR glasses and AI glasses lies not only in display capabilities but also in interaction methods. AI glasses use VUI (Voice User Interface), primarily relying on voice, TTS (Text-To-Speech), and indicator lights for feedback interaction; whereas AR glasses employ GUI (Graphical User Interface), integrating display, touch control, voice, and TTS for comprehensive interaction.

However, both AR glasses and AI glasses confront the 'impossible triangle' challenge of battery life, weight, and computing power. Like all smart terminals, the more functions, the higher the energy consumption, and expanding battery capacity increases the weight of the glasses. This is particularly evident in AR glasses, whose cumbersome headset design deters many potential users. Although AI glasses without a display screen reduce weight to some extent, compared to the lightweight 20-30 grams of ordinary glasses, their approximate 45 grams (without lenses) still feel heavy, and the weight increases further with lenses, making them uncomfortable for extended wear.

Moreover, battery life is a significant concern for AI glasses. Compared to handheld devices like phones and cameras, the continuous recording time of AI glasses like Meta is only about 3 hours at most, far from meeting users' needs for prolonged use.

This indicates that smart glasses products are still in a nascent stage. Both AR glasses and AI glasses have numerous technical hurdles to overcome before the industry truly explodes. Notably, AR glasses are actively integrating AI technology to expand application scenarios, while some AI glasses manufacturers are striving to enhance display capabilities to compensate for their shortcomings.

Behind this trend lies the industry's universal acknowledgment of AR glasses as the ultimate form of smart glasses. Professionals highlight that over 80% of human information is received visually; without visual interaction, relying solely on AI capabilities is challenging to become a new entry point for humans to obtain information and services. Xu Chi, founder and CEO of XREAL, also stated that AR is the best carrier for AI, and AI is the optimal interaction method for AR. These two technologies are poised to be the core technologies that will most transform consumer electronics and terminal experiences in the next decade.

Looking ahead, AR and AI technologies in smart glasses may converge. However, for now, both face the challenge of achieving large-scale, rapid market penetration. After all, even the industry hit product Ray-Ban Meta only sold over a million units in a year, a number that for mobile phone manufacturers is merely a few days' sales after a new product launch.

In the short term, with a relatively vague user profile for AR glasses, smart audio glasses aim to replace earphone functionality, using glasses (for vision correction/decoration) as the carrier and enhancing the experience and selling points with AI, potentially opening up a new audio segment.

Amidst the 'Hundred Glasses Competition,' it's currently challenging to determine a winner, so institutional investors are still focusing on manufacturers entering various product supply chains or possessing relevant technological reserves. These include:

Optics: Crystal Opto-Electronics, GoerTek, O-Film Tech, Weltrend Semiconductor

Assembly: GoerTek, Edify Information, Tianjian, Longqi Technology, Jiahe Intelligence

Storage: GigaDevice Semiconductor

Camera: Weltrend Semiconductor, Sunny Optical Technology (Group) Company Limited

SoC: Thine Electronics, Rockchip, Bluecore, Actions Semiconductor

Structural components: JMT, Changying Precision Technology

Screens: BOE Technology Group Co., Ltd., HC SemiTek, JBD (unlisted)

Terminal manufacturers: Rokid (unlisted), Leiniao Innovations (unlisted), Xreal (unlisted), INMO (unlisted).

- End -