CITIC Securities: Volcano Engine is empowering the AI implementation of multiple categories of hardware products, focusing on Byte Ecological Chain

The Zhitong Finance App learned that CITIC Securities released a research report saying that on June 11, ByteDance's Volcano Engine hosted the Force2025 Propulsion Conference. In addition to the main forum, Volcano Engine also held product-specific forums, industry-specific forums, and partner forums. Through this conference, I saw that Volcano Engine is empowering the implementation of AI in multiple categories of hardware products, and that the application of multi-modal visual understanding is a clear trend. Looking forward to the future, I am optimistic that AI will be widely implemented on the hardware side, and AI glasses and AI toys are expected to become the most typical products. It is recommended to focus on Byte Ecological Chain and parts and module companies that bind the growth of brand customers.

CITIC Securities's main views are as follows:

Multiple categories of bytes empower AIoT and focus on multi-modal implementation.

I saw smart hardware products such as AI alarms, AI learning machines, AI toys, AI mice, AI glasses, AI watches, AI headphones, AI mattresses, AI pillows, AI security cameras, AI PCs, AI phones, AI pets, robots, etc. at the Volcano Engine Force 2025 Power Conference. The big model is expanding into many categories on the hardware side. Mr. Xing Xiaoci, the person in charge of intelligent hardware for the Volcano Engine Big Model, introduced the advantages of the Volcano Big Model using AI headsets and AI toys as examples. The specific example is that AI toys can be easily configured with characters, sounds, and motion controls. AI headsets return an average of 1.5s based on the ASR+ bean bag model, and they can all achieve rapid turnkey delivery and landing. According to Mr. Xing Xiaoci, as of June 11, the number of AIoT products connected to bean bags had been shipped in excess of 1 million units (note: in a follow-up speech, Mr. Gao Feng, the leading AI toy company in China, said that the company's shipments account for about a quarter of this). Mr. Xing Xiaoci expects this figure to break 10 million units by the end of this year.

Judging from the trend of implementation, multi-modality incorporating visual understanding is the focus of byte layout. Mr. Wu Di, the person in charge of Volcano Engine's intelligent algorithm, gave two examples of hardware implementation: 1) With the power of the big model, a security camera is not only a camera, but also a housekeeper and personal assistant. 2) Equipped with a camera, the desk lamp can be used as a learning machine to take pictures and solve problems, and with a printer, it can help students sort out a collection of wrong questions.

Industrial chain companies participated in the event. Among them, Broadcom Integrated, Xingchen Technology, and Rokid delivered keynote speeches.

At the conference, I saw companies with a very multi-byte ecosystem, such as Zhongke Lanxun, Yiyuan Communications, Xingchen Technology, and Broadcom Integration. Among them, Mr. Zhang Pengfei, Chairman of Broadcom Integration, and Mr. Chen Lijing, Vice President of Xingchen Technology, were invited to deliver speeches at the Smart Hardware Sub-Forum. Mr. Zhang Pengfei focused on the advantages of low latency, high bandwidth, and ultra-low power consumption, highlighting that chip products are environmentally optimized end-side AI experiences in Doubao; Mr. Chen Lijing focused on multi-modal applications. The company has now deployed chip products in homes, businesses, and wearables (such as glasses). In addition, Mr. Wang Junjie, vice president of Rokid, a famous domestic eyewear brand company, delivered a speech focusing on multi-modal applications in eyeglasses; Mr. Gao Feng, a leading domestic AI toy company, focused on end-to-end toy products, and stated that the company is expected to release the world's first end-to-end AI toy within the year.

As the release date of the Xiaomi Glasses approaches, sentiment on the end side is expected to pick up.

According to the WeChat account XR Vision, Xiaomi Glasses is expected to be held on June 16 and released on June 26. The product uses a Qualcomm AR1+ Hengxuan 2700 chip configuration and is equipped with a Sony IMX 681 image sensor. Xiaomi was the first mobile phone manufacturer to release AI glasses. Considering the product's exposure and fan stickiness, there is a possibility that the product's subsequent sales tracking will exceed expectations, and investors are advised to keep a close eye on it. Catalyzed by this incident, I am optimistic that the end-side sector represented by AI glasses will pick up sentiment. In addition to AI glasses in the form of Rayban META, it is emphasized that investment in the eyewear sector in 2025 should consider the evolution of the product form, that is, iterating from glasses without a display to glasses with a single green display and full color display. Therefore, it is recommended to pay attention to the pace of release of related products from manufacturers such as META, Rokid, and Video; the supply chain suggests focusing on aspects such as optical waveguides and MicroLED.

Risk factors:

Downstream demand falls short of expectations; product innovation is slow; product application scenarios are limited; rising terminal hardware costs affect demand; changes in the international industrial environment and trade frictions intensify; AI commercialization progress falls short of expectations; and large-scale model iterations fall short of expectations.