By 2024, AI on mobile devices is expected to experience rapid development. With the advancement of AI large-scale model technology and the improvement of mobile AI chip computing power, it has become possible to run local large-scale models on mobile devices. Samsung, Oppo, Vivo, Xiaomi, Glory, Red Magic and other domestic and foreign mobile phone manufacturers have also successively released or planned to release mobile terminal products and corresponding apps with generative AI capabilities. The landing of large-scale AI models in mobile phones will inject new power into the smartphone manufacturing industry, and the daily and high-frequency application of AI will put higher demands on mobile hardware and components, driving the iteration of mobile hardware and benefiting mobile component suppliers.
According to the report released by Dongfang Securities, Apple may release iOS 18 at WWDC, which might integrate AI features from OpenAI and Google Gemini. By 2024, AI on mobile devices is expected to experience rapid development. With the advancement of AI large-scale model technology and the improvement of mobile AI chip computing power, it has become possible to run local large-scale models on mobile devices. Samsung, Oppo, Vivo, Xiaomi, Glory, Red Magic and other domestic and foreign mobile phone manufacturers have also successively released or planned to release mobile terminal products and corresponding apps with generative AI capabilities. The landing of large-scale AI models in mobile phones will inject new power into the smartphone manufacturing industry, and the daily and high-frequency application of AI will put higher demands on mobile hardware and components, driving the iteration of mobile hardware and benefiting mobile component suppliers.
Orient Securities' main points are as follows:
Large-scale AI models are expected to reshape mobile phone functions, and the basic landing foundation for terminal large-scale models is ready.
Generative AI enables mobile phones to create completely new digital videos, images, texts, audios or codes, reshaping AI mobile phone functions. Thanks to the breakthroughs in large-scale models and the decreasing cost of terminal computing, domestic and foreign mobile phone manufacturers have successively released self-developed AI large-scale models and corresponding mobile operating systems as well as app upgrades. With the advancements in upstream chip computing power, the chip basis for terminal-side large-scale models is already in place. Since September 2023, Apple, Google, MediaTek and Qualcomm have all released new generations of high-performance smartphone chips that can support the local operation of large-scale AI models with parameters of over one billion.
AI enriches the application scenarios of smart phones and promotes the iteration of abilities in image capture, human-computer interaction, identity authentication and system optimization, etc. With the support of AI algorithms and computing power, mobile phones' image capturing ability in typical scenarios such as portrait, telephoto and low light has been enhanced. Generative AI allows mobile phones to adjust shooting content and even achieve virtual-real combination. Language models were the first to be landed, and voice assistants and text processing will become standard features of flagship smartphones. AI can enhance the perception and management of internal resources of mobile phones, optimize chip performance and promote the iteration of mobile hardware. 5.5G will bring ten times the network performance improvement, and ultra-reliable and ultra-low latency communication will improve user experience, enrich vertical application scenarios and promote end-side intelligence as well.
AI driving industrial innovation will stimulate the growth of the mobile phone chip market and intensify competition in the mobile phone chip market. The application of large-scale AI models drives demands for the computing power, power consumption and cost of mobile terminal chips, and major chip manufacturers such as Qualcomm and MediaTek have successively released new generations of chips for terminal-side large-scale AI models, achieving significant improvements in AI chip computing power. AI promotes the intelligent development of mobile phone sensors, and the development and popularization of AI applications such as image recognition and mobile assistants will accelerate the demand for multimodal human-computer interaction, promoting the transformation of sensor components to more intelligent directions. The demand for 3D intelligent cameras and various bio-sensors (e.g. fingerprint and iris sensors) is expected to increase.
AI is expected to promote the progress of the mobile phone industry chain.
AI big model applications drive the demand for mobile terminal chip computing power, power consumption, and cost, stimulating the growth of the mobile phone chip market and intensifying the competition. Major chip manufacturers, such as Qualcomm and MediaTek, have successively released new generations of chips for end-side AI big models, thus achieving a significant improvement in AI chip computing power. Besides, AI promotes the intelligent development of mobile phone sensor components, and the development and popularization of AI applications such as AI images and mobile assistants will accelerate the demand for multimodal man-machine interaction, thus promoting the transformation of sensor components to more intelligent directions.
AI will promote the transformation of operating systems and change the software ecology of mobile phones. AI promotes the important strategic transformation of mobile phone operating systems, making them develop into intelligent assistants and electronic advisors; end-side AI makes it possible for AIApps to run locally, which may be widely welcomed due to the needs of low latency and data security.
Target of attention:
Chips: SMIC (688981.SH), TianShui Huatian Technology (002185.SZ), JCET Group (600584.SH), TongFu Microelectronics (002156.SZ).
Storage: GigaDevice Semiconductor Inc. (603986.SH), Dongxin Shares (688110.SH), Hengshuo Shares (688416.SH), Gaintech Semiconductor Corporation (688123.SH), Purui Technology (688766.SH), Jiangbolong (301308.SZ), Deming Li (001309.SZ), Netac Technology (300042.SZ), Barefoot Storage (688525.SH), Shenzhen Kaifa Technology (000021.SZ).
Cooling: Jones Tech PLC (300684.SZ), Carbon Yuan Technology, Fullhan Microelectronics Co. Ltd(688662.SH), Hongfuhan Technology Inc.(301086.SZ), Shenzhen FRD Science & Technology Co., Ltd. (300602.SZ).
Battery life: Zhuhai Guanyu (688772.SH), Sunwoda Electronic Co., Ltd. (300207.SZ), Haopeng Technology Co., Ltd. (001283.SZ), Shenzhen Desay Battery Technology Co., Ltd. (000049.SZ), Shenzhen Zijian Electronics Co., Ltd. (301121.SZ), Xiamen Tungsten New Energy (688778.SH), Lingyi Itech (002600.SZ), Anker Innovations Technology (300866.SZ).
Miniaturization of components: Shenzhen Sunlord Electronics (002138.SZ), Guangdong Faith Long Crystal Technology (300460.SZ), Universal Scientific Industrial (Shanghai) Co., Ltd. (601231.SH), JCET Group Co., Ltd. (600584.SH), Luxshare Precision Industry (002475.SZ), Avary Holding (002938.SZ), Suzhou Dongshan Precision Manufacturing (002384.SZ), Shenzhen Everwin Precision Technology (300115.SZ), Shenzhen Hui Chuang Da Technology (300909.SZ), Hongruida Technology (301285.SZ).
Sensors: Sunny Optical Technology (02382), OFILM Group Co., Ltd. (002456.SZ), Q Tech (01478), Cowell (01415), Will Semiconductor (603501.SH), Zhejiang Crystal-Optech Co., Ltd. (002273.SZ), Zhejiang Lante Optics Co., Ltd. (688127.SH), Shenzhen Goodix Technology Co., Ltd. (603160.SH).
Risk warning: The demand of downstream mobile phone market is lower than expected; the progress of large AI models in end-side is lower than expected; the upgrading of mobile phone hardware technology is lower than expected; the results may be affected by changes in assumptions.