华金证券：Mobile phone manufacturers are seizing the AI wave, and a new round of smartphone hardware demand is also rising.

Zhitong App learned that Huajin Securities has released a research report stating that smartphones, as important end devices, are significantly benefiting from the comprehensive user experience upgrade brought by AI large models, which may stimulate a new round of consumer demand. Smartphone manufacturers are entering the market one after another, seizing the AI smartphone trend. According to The Economic Times in March 2024, Apple's iOS 18 is expected to be launched in September 2024, with the generative AI large model expected to be integrated into Siri. In addition, domestic manufacturers such as Xiaomi, VIVO, OPPO, and Meizu have also successively entered the field of AI smartphones. With more and more manufacturers accelerating their entry, a new round of demand for smartphone hardware is also rising. Huajin Securities' views are as follows: The trend of mixed AI is emerging, smartphones significantly benefit from the rise of end-side AI The hybrid AI that works collaboratively with cloud AI and end-side AI can achieve a better application experience and more efficient resource utilization, which is the future development trend. Influenced by the global economic downturn and the innovation bottleneck of smartphones, routine updates of software and hardware specifications are difficult to attract consumers. Smartphones, as important end devices, are significantly benefiting from the comprehensive user experience upgrade brought by AI large models, which may stimulate a new round of consumer demand. Qualcomm stated that smartphones can currently run AI large models with parameters exceeding 10 billion, and in the near future, they are expected to run large models with parameters reaching a trillion. Hardware empowers end-side AI operation, Qualcomm and MediaTek take the lead Qualcomm, MediaTek, and other manufacturers have successively launched mobile SoC chips with powerful AI computing power to provide underlying performance and technical support for end-side generative AI. In October 2023, Qualcomm officially released the first chip designed for mobile devices for generative AI - Snapdragon 8 Gen 3. This chip adopts a fusion AI acceleration architecture, upgrading Hexagon DSP to Hexagon NPU. The overall performance of Hexagon NPU has been improved by up to 98%, equipped with an independent power supply circuit, and the energy efficiency has increased by about 40%. MediaTek stated that compared with typical CPUs and GPUs, its APU can provide up to 27 times and 15 times power efficiency for CPUs and GPUs, respectively. Smartphone manufacturers are entering the market one after another, seizing the AI smartphone trend In November 2023, Samsung released its self-developed generative AI product Gauss, involving three major functions: text processing, code writing, and image generation. In January 2024, Samsung launched the Galaxy S24 series at a global new product launch event with the theme "Opening a New Era of Mobile AI." Samsung stated that within just 28 days of its launch, the Galaxy S24 series had already exceeded one million units in sales in South Korea. In December 2023, Apple released two papers on AI large models, introducing new technologies for 3D virtual characters (Avatar) and efficient language model inference. According to Bloomberg in February 2024, Apple has decided to abandon its electric car project and transfer some of the electric car project employees to the AI department. According to The Economic Times in March 2024, Apple's iOS 18 is expected to be launched in September 2024, with the possibility of integrating generative AI large models into Siri. In addition, domestic manufacturers such as Xiaomi, VIVO, OPPO, Meizu, and others have also entered the AI smartphone field one after another. With more and more manufacturers accelerating their entry, a new round of demand for smartphone hardware is also rising. Risk Warning: If the demand for smartphones does not recover as expected, the hardware technology research and development process falls short of expectations, the application effect of large AI models on the mobile side is below expectations, market competition intensifies, systemic risks, etc.