2024 is known as the first year of AI phones, and the Samsung Galaxy S24 is selling well in South Korea, breaking the sales record for this model by surpassing 1 million units in just 28 days. This shows that the application of AI technology will accelerate the arrival of the industry's replacement wave, and mobile phone sales may usher in an explosion. In contrast, Apple's iPhone 15 has seen a sharp decline in sales since its release in September 2023, and in the face of decline, Apple has abandoned its ambitious plans to build a car and fully turned to the AI battlefield.
The AI capabilities of the Samsung Galaxy S24 are mainly used in areas such as translation, search, and imaging. Translation features include real-time translation and simultaneous interpretation to provide users with more efficient services. The search function focuses on "immediate circle search", which changes the traditional search method. Imaging AI capabilities can intelligently fill** to improve the user experience.
AI mobile phones have set off a new wave in the market. Samsung fired the first shot of AI mobile phones, and major brands have followed suit
Samsung fired the first shot of mobile AI in 2024, Google model blessed galaxys24, and a new search debut, which can be searched by drawing a circle. The next-generation flagship galaxy S24 is the first smartphone to be equipped with a large model GeminiPro and the powerful ImageN2 on Google Cloud Platform VertexAI. In addition to the Samsung brand, at the recently held Mobile World Congress (MWC2024) with the theme of "FutureFirst", Honor announced that it will join hands with Qualcomm and Meta to introduce a 7 billion parameter large model to the device side. OPPO has also officially announced that it will fully focus on the AI field, and will also introduce a series of generative AI features in the second quarter of 2024, including AI elimination. In addition, on the evening of February 29th, Meizu's first open AI terminal Meizu 21Pro was officially released, Meizu 21Pro can realize a number of AI functions including AI voice assistant and AI gallery, which can be used for professional knowledge Q&A, and can also generate text or ** according to natural language, and can also help users automatically generate message replies and create long texts according to user needs. At the same time, Huawei's HarmonyOS4 system is fully connected to the Pangu model, and the "smart map" function based on multi-modal model technology has appeared in flagship mobile phones, and the emergence of AI mobile phones has changed people's lives. In the future, the AI capabilities of Huawei Mate70, iPhone 16 and other products are the most expected, and the competition of multi-brand AI mobile phones has begun to promote the upgrading of mobile devices.
Mobile phone shipments are declining, and AI phones are expected to reverse the decline
According to a report published by IDC, global smartphone shipments in 2023 hit the lowest in 10 years, with an increase of only 8% in the fourth quarter, and the annual mobile phone shipments were 11700 million units, a year-on-year decrease of 32%。In the future, the replacement of mobile phones driven by AI phones is expected to reverse this decline, and China, as an emerging market, is also an important driving force for mobile phone growth.
According to IDC data, the global shipment of next-generation AI mobile phones in 2023 will be 05.1 billion units, with shipments expected to reach 1700 million units, accounting for about 15% of overall smartphone shipments; Although the penetration rate of AI phones in the Chinese market in 2024 is lower than that of the world (132%), but the share will climb rapidly after 2024, and China's AI phone sales are expected to reach 1500 million units, with a market share of more than 50%. Figure 3 (IDC **AI Mobile Phone Penetration).In the future, with the acceleration of the penetration rate of AI mobile phones, the consumer electronics industry chain will usher in a new cycle.
AI technology has become a trend, leading a new round of hardware innovation
Recently, a new technology product was born - NPU, that is, the neural network processing unit, which can simulate human neurons and synapses at the circuit layer, and directly process large-scale neurons and synapses with a deep learning instruction set, and complete the processing of a group of neurons with one instruction. Compared to the von Neumann structure of CPUs and GPUs, NPUs integrate storage and compute through synaptic weighting, thereby improving operational efficiency.
NPU is a type of ASIC chip, which is currently mainly used in artificial intelligence tasks such as deep learning and machine learning, and unlike the frequent interaction between GPUs and CPUs, NPU-accelerated neural network computing efficiency is more efficient, and will also be widely used in AI devices in the future.
Taking the Snapdragon 8Gen3 as an example, the Qualcomm AI engine used has a powerful NPU for mobile devices, integrating upgraded hardware acceleration units, micro-slice inference units, enhanced tensors, scalar and vector units, all of which share a large shared memory with 2x the bandwidth. Mixed precision of INT8+INT16 and all accuracy of INT4, INT8, INT16, and FP16 are supported. According to the Heart of the Machine, it delivers 98% better performance and 40% more energy efficiency than its predecessor. For the first time, the Snapdragon 8gen3 supports models running 10 billion parameters, and the scale has reached the scale of 10 billion. At the same time, the time to generate an image has been reduced to less than 1 second, which is the fastest speed. When running the Meta large language model LLAMA2-7B, Snapdragon 8Gen3 can generate 20 tokens per second, which is also one of the fastest on the mobile phone terminal. In addition to the birth of new hardware technologies, traditional electronic devices will also need to be updated in AI mobile phones.
The computing power of large models requires sufficient memory, and technology updates promote product iteration
The implementation of large models requires not only the cooperation of computing power, but also sufficient memory space for deployment. At present, the large models implemented on the device side are mainly based on billions of parameters, and higher-specification models are difficult to commercialize in the short term because they require more memory usage. In terms of solving the path, in the case of highly integrated NPUs, it is necessary to increase the memory configuration of the terminal itself. In the future, if the NPU is as independent as the GPU, its video memory needs to be further developed to balance the storage, transmission, and computing power of the chip, so as to provide all-round support for the terminal model to run in an ideal state. Taking vivo as an example, a new technology path has been formed. In the future, new storage products will also face upgrading, which will drive industry demand.
The amplification of computing power has led to an increase in power consumption, and the market for heat dissipation devices has expanded
Heat dissipation capability is also a major bottleneck for high-computing power chips, and with the continuous improvement of model specifications mounted on AIPC and AI phones, the release of NPU performance may become more aggressive. At present, in terms of consumer electronics terminals and mobile phones, stainless steel VC vapor chambers have gradually replaced copper VC vapor chambers as the main force for heat dissipation; The heat dissipation scheme of a PC is generally presented in the form of a combination, while the heat dissipation module mainly includes a heat dissipation base, heat pipes, heat fins, cooling fans and other objects from a structural point of view.
Investment Advice:
At present, in the context of artificial intelligence, major industries have begun to move closer to it, Samsung is the first brand to launch AI mobile phones this year, which has driven the market heat, and the current market is expected to be born in Apple and Huawei and other manufacturers' new AI mobile phones, if the function also exceeds expectations, it is expected to drive the second wave of heat in the market, especially before and after this year's mobile phone conference, the concept of AI mobile phones should be focused on tracking.
It is recommended to pay attention to:AI phones, AIPCs, consumer electronics, electronic devices.
002632 TD Optics
300684 Zhongshi Technology
002579 Zhongjing Electronics
603160 Goodix
300647 Overclocking III
301489 Siquan New Materials
Risk Warning:
The market is risky, and investors need to be cautious. Shanghai Huizheng Financial Consultant is an investment consulting company approved by the China Securities Regulatory Commission, with a unified credit code of 91310107MA1G0KQW5N. This article is for investor education purposes only and does not constitute any investment advice, and investors act accordingly at their own risk. The information referred to in this article is accurate and reliable, but does not make any guarantee for its accuracy, completeness and timeliness, and does not assume any responsibility for all losses caused by investors' investment based on it.