生成式AI同智能终端结合，具备三点优势：①个人信息无需上传云端，隐私泄露和数据安全风险降低;②AI模型接入本地数据库和个人信息，有望实现通用基础AI大模型向个性化定制小模型转变，提供更适合的用户服务;③通过压缩AI大模型和终端软硬件适配，边缘AI可能降低运行成本、加快响应速度和提高服务效率。因此，联想、谷歌和微软均发布AI PC或AI手机等产品，小米14将成为首发搭载高通骁龙8 Gen 3处理器的手机;软件端将AI大模型嵌入终端设备，硬件端提升AI相关硬件的性能，帮助智能终端搭载的聊天机器人、实时翻译、AI相机、生产工具等AI功能的种类完善和性能增强。The Zhitong Finance app learned that Everbright Securities released a research report saying that edge AI has the advantages of security, customization, and efficiency, and that AI PCs and AI phones are expected to become a new trend in consumer electronics transformation. Recently, technology vendors such as Lenovo Group (00992), Qualcomm (QCOM.US), Intel (NVDA.US), and Google (GOOGL.US) have released edge AI smart terminal products and AI processors, and the wave of generative AI has spread from the cloud to the edge. The bank said that considering the extension of AI computing power from the cloud to the edge, there is considerable market space for AI chip manufacturers. Furthermore,AI smart terminals such as AI PCs and AI mobile phones may become the direction of change in the consumer electronics industry in the AI era in the future, and the pioneering layout of hardware manufacturers is conducive to determining the direction of innovation and market position.
The main views of Everbright Securities are as follows:
Smart terminals: Edge AI has the advantages of security, customization and efficiency, and AI PCs and AI phones are expected to become a new trend in consumer electronics transformation.
Combined with smart terminals, generative AI has three advantages: ① no need to upload personal information to the cloud, privacy leakage and data security risks are reduced; ② The AI model is connected to local databases and personal information, which is expected to transform the general basic AI model to a personalized small model to provide more suitable user services; ③ By compressing the large AI model and the adaptation of terminal software and hardware, edge AI may reduce operating costs, speed up response speed, and improve service efficiency. As a result, Lenovo, Google, and Microsoft have all released products such as AI PCs or AI phones, and Xiaomi 14 will be the first mobile phone equipped with a Qualcomm Snapdragon 8 Gen 3 processor; the software side will embed a large AI model into the terminal device, and the hardware side will improve the performance of AI-related hardware to help improve the variety and performance of AI functions such as chatbots, real-time translation, AI cameras, and production tools on smart terminals.
AI processor: The combination of CPU+GPU+NPU is favored, and chip manufacturers focus on energy efficiency and AI performance.
1) Chip manufacturers' design goals are shifting from simply pursuing performance to energy efficiency and AI performance. Qualcomm released the Snapdragon X Elite processor for PCs. The AI computing power reached 45 TOPS (4.5 times the speed of its peers) and reduced power consumption by 65% compared to the Intel Core i7-13800H at peak performance; Qualcomm released the Snapdragon 8 Gen 3 platform for mobile phones, achieving a 30% and 25% increase in CPU and GPU energy efficiency, respectively, and a 98% increase in NPU's AI performance. Intel launched the Meteor Lake processor to effectively improve energy efficiency and AI performance through a “separate module architecture” and “CPU+GPU+NPU flexible scheduling.” 2) Under the combination of CPU+GPU+NPU, the Arm architecture is expected to increase its market share in the PC field due to its low power consumption, excellent performance, and long battery life. Both Nvidia and AMD plan to develop PC chips based on ARM architecture, which are expected to be released in '25.
Investment advice:1) Considering the expansion of AI computing power from the cloud to the edge, there is considerable market space for AI chip manufacturers. I recommend Nvidia, and focus on AMD, Qualcomm, and Intel; 2) AI smart terminals such as AI PCs and AI phones may become the future direction of change in the consumer electronics industry in the AI era. The pioneering layout of hardware manufacturers is conducive to determining the direction of innovation and market position. I recommend Lenovo Group. It is recommended that attention be paid to Google, Microsoft (MSFT.US), and Xiaomi Group-W (01810); 3) Arm is expected to take advantage of the edge AI to increase its market share in the PC market. It is recommended to focus on ARM (ARM.US).
Risk analysis:Demand recovery in the consumer electronics market falls short of expectations; AI commercialization progress falls short of expectations; edge AI performance falls short of expectations; domestic and foreign policy risks.
