In 2024, the AI wave swept the globe, and the consumer electronics industry also tried to take advantage of this opportunity to achieve a revival.
In the past year, the leapfrog development of generative artificial intelligence and large models has attracted global attention. In the consumer electronics industry, large models of artificial intelligence are driving a new round of technological innovation at the edge and end of consumer electronic products.
Industry insiders believe that the application scenarios of artificial intelligence have innovated the experience of human-computer interaction at the end of consumer electronics, accelerating the process of electronic and intelligent transformation of various terminals, which will have a huge impact on the consumer electronics industry chain.
Advertisement
George Miao, Chairman of Qualcomm China, said that artificial intelligence, especially generative artificial intelligence, brings "not a gradual improvement", but an unprecedented technological revolution, and smartphones will be one of the fastest developing fields of generative artificial intelligence. In the next one or two years, more generative artificial intelligence technology will be implemented in consumer electronic terminals such as personal computers and smartphones, making human-computer interaction more convenient and humanized. It only needs text or voice input to help users quickly complete tasks and meet various needs. This is a completely different experience for consumers compared to the past.
01
The combination of AI and consumer electronics
Artificial intelligence refers to the technology of simulating human intelligence using computer science and large data sets, including machine learning, deep learning, natural language processing, etc. Consumer electronics are electronic products for the general consumer, including smartphones, tablets, smart home devices, wearable devices, etc.
The combination point of AI and consumer electronics lies in that AI technology can endow consumer electronic products with more intelligent functions and more humanized experiences. For example:
AI smartphones: Compared with other end-side devices, mobile phones have four significant advantages of high popularity, strong computing power, rich interactive and application scenarios, and high stickiness, making them more suitable as carriers of AI technology. Generative AI is empowering smartphones in all aspects, turning them into all-weather private intelligent assistants and mobile productivity tools.The attempt to empower smartphones with AI technology can be traced back to as early as 2017, when Android phones integrated a dedicated AI computing unit in their SOC platform to run and influence the enhancement of related deep learning models. Subsequently, AI technology has gradually been applied in areas such as strengthening security, optimizing battery life, and improving network performance, but computational photography remains the most important application field. It was not until Large Language Models (LLMs) were installed in smartphones that mobile AI applications entered the era of large models. Currently, top Android phone manufacturers such as VIVO, OPPO, and Honor have successfully achieved the local deployment of 7 billion parameter large models.
Generally speaking, generative AI phones utilize large-scale, pre-trained generative AI models to achieve multimodal content generation, situational awareness, and possess continuously enhanced human-like capabilities. Generative AI phones need to have the following four essential characteristics:
Support for the local deployment of large models, or the execution of complex generative AI tasks through cloud collaboration. Generative AI phones themselves have strong AI computing power and do not need to rely entirely on cloud servers.
Possess multimodal capabilities, that is, they can process various forms of content input such as text, images, and speech to generate various forms of output, with typical examples including translation, image generation, and video generation.
Ensure a smooth and seamless user experience, where the phone can interact naturally and intuitively, quickly responding to user requests.
Have hardware specifications to achieve the above three points, including but not limited to mobile computing platforms designed based on leading processes and advanced architectures, with integrated or independent neural network computing units (such as APU/NPU/TPU), large capacity and high bandwidth memory, and stable and high-speed connections, as well as hardware-level and system-level security defenses.
AI PC: that is, a PC with an integrated AI engine, capable of AI perception of voice and image information, while the software system is compatible with AI applications such as AI search, content generation, intelligent recommendations, and can contribute AI computing power. Under the wave of AIPC, the experience of related application scenarios has been greatly upgraded.
On May 21, 2024, Microsoft launched a new brand of Windows PC - Copilot+PC, which is a Windows PC with an NPU core and Microsoft AI assistant Copilot. Compared with the previously launched AIPCs, Microsoft has fully introduced its AI assistant Copilot into the Windows system this time, and has built-in the OpenAI GPT-4o model, which can directly process artificial intelligence accelerated workloads (such as running AI chatbots and image generation) locally without relying on the cloud. According to Microsoft's definition, the hardware configuration of Copilot+PC requires at least 16GB of memory (DDR5/LPDDR5), 256GB SSD/UFS, and the performance of the integrated NPU needs to reach more than 40 TOPS. Microsoft also announced a cooperation with top OEMs such as Acer, ASUS, Dell, HP Lenovo, Samsung, etc., to launch a new Copilot+PC product starting on June 18.
Under the wave of AI PCs, the experience of related application scenarios has been greatly upgraded:
With the continuous advancement of the Internet and communication technology, mobile office work is expected to become the future work model. AI technology can optimize data transmission schemes to enhance real-time collaboration and strengthen audio and video effects through AI noise reduction technology, ultimately bridging the sensory perception differences between remote interaction scenarios and face-to-face scenarios.AI technology can accelerate the processing speed of image-based information, empowering the content creation market in game development, graphic design, photography, short video, and more. Content creation through AI requires stable and efficient AI computing power support from terminal devices.
With the support of AI technology, terminal devices are becoming increasingly powerful in performance and functionality, assisting users in handling daily information with more humanized and customized interaction modes, and better serving as personal digital assistants.
AI PCs can optimize the visual effects and sound effects of games without increasing the burden on the GPU, while greatly enhancing the interaction experience between players and virtual characters in the game.
AI TV: In March of this year, Samsung started selling the 2024 Neo QLED 8K TV. This TV is equipped with a "third-generation AI 8K processor" with 512 neural networks, eight times that of the previous generation, and the neural processing unit (NPU) speed is twice as fast as the previous generation. It can instantly analyze and magnify content, allowing viewers to clearly enjoy low-resolution videos, and can detect volume differences in each piece of content, separate and amplify sounds to ensure that dialogue is clearly conveyed without being drowned out by background noise or external sounds.
In addition to Samsung, Hisense also released a new TV model U8N Pro in March, equipped with a Xinxin AI perception chip, and the image enhancement capability has also been improved. According to the official statement, it has fully entered the era of AI computing picture quality. In May, Hisense released the AI Mini LED TV new product X Ultra series, which uses Hisense's Xinghai large model.
AI TVs are different from traditional smart TVs. Traditional smart TVs are TVs with a fully open platform and an operating system, allowing users to install various application software. AI TVs, on the basis of smart TVs, combine AI technology to make the functions of TVs more powerful. With AI technology, it is also possible to achieve a comprehensive upgrade of TV audio and video quality.
In 2024, AI TVs have just hit the market. Currently, there is no unified definition of AI TVs, and they are still exploring the path of development. The combination of "AI + TV" does not seem to have an ultimate form. From the products that have been released, the core elements of AI TVs can be summarized as: AI picture quality, AI interaction, AI interconnection, and AI creation.
It is not difficult to see that the main integration points of AI and consumer electronics are AI imaging technology, AI voice assistants, AI interaction, AI large models, and AI chips such as NPU.
02
Major manufacturers fully embrace AICurrently, many PC, smartphone, and even TV manufacturers are using AI as a key differentiating factor, hoping to develop new and eye-catching features through AI. They have even proclaimed the slogan "All in AI," fully embracing AI technology.
Since 2017, the smartphone industry has entered a stock market, and the innovation of smartphones has entered a bottleneck period. Smartphone manufacturers have started to compete in configuration and price, and AI large models have brought new opportunities to smartphone manufacturers. According to Canalys' forecast, AI smartphones will account for 5% of the total shipment of the global smartphone market in 2024, and this proportion will rise to 45% by 2027. For the Chinese market, IDC predicts that the share of AI smartphones will rise rapidly after 2024, reaching 150 million units in 2027, with a market share of over 50%.
Major smartphone manufacturers such as VIVO, OPPO, Honor, Huawei, and Xiaomi have all released AI large models, evolving towards generative AI smartphones. In February 2024, Meizu decided to go all in on AI and stopped new projects for traditional smartphones; OPPO announced the official entry into the era of AI smartphones, and on New Year's Eve, it pushed more than 100 AI practical functions to more than 10 smartphone series. Apple also announced at the end of February 2024 that it would give up car manufacturing and turn to generative AI, and the next-generation iPhone 16 may become Apple's first AI smartphone. Huawei also announced that HarmonyOS NEXT will deeply integrate AI with the operating system, building a new Hongmeng native intelligent framework. Not long ago, Xiaomi announced that the Xiaomi Xiaoai large model will be fully upgraded and all free. In addition to smartphones and tablets, devices such as TVs, cars, and speakers will also support upgrades.
Some smartphone manufacturers' AI large models on smartphones
It is worth noting that the NPU is the core hardware to achieve smartphone AI functions. In addition, the embedding of large models has put forward higher requirements for the memory capacity, bandwidth, and transmission rate of smartphones. In the future, 16GB may become the basic configuration of smartphone memory. These smartphones equipped with SoCs that can run generative AI models on the side and have NPU computing power greater than 30 TOPS with int-8 data types are also called "new generation AI smartphones." Products that meet the above SoC definition include Apple A17 Pro, MediaTek Tianji 9300, and Qualcomm Snapdragon 8Gen3.
In addition to AI smartphones, many technology giants are also fully betting on AIPC. In the field of AIPC, manufacturers including NVIDIA, Intel, Lenovo, HP, and Acer are all accelerating research and innovation, hoping to occupy a place in the new field. It is expected that AI PCs will usher in the initial stage of volume in 2024, become mainstream products in the market in 2025, account for more than half of the overall PC sales in 2026, and the shipment proportion will rise to 60% in 2027.
Lenovo previously released the industry's first "AI PC Industry (China) White Paper," which comprehensively revealed the future of the AI PC industry, stating that it will continue to bet on the direction of AI PCs in the future. Now, Lenovo is also cooperating with the technology giant Alibaba Group and has launched an AI computer based on Alibaba's large language model. In addition, Lenovo also plans to equip all new PCs with artificial intelligence (AI) functions by 2027.
At the beginning of this year, Microsoft, with the advantages of operating system and AI, personally entered the market and released its own AI PC - Surface Pro10 commercial version and Surface Laptop6 commercial version. Apple also released a new MacBook Air, which Apple claims to be the world's best AI consumer-level notebook computer.
At present, chip processor manufacturers such as Intel, Qualcomm, and AMD have launched processors for AI PCs, and brand manufacturers such as Lenovo, ASUS, Dell, and Honor are also actively promoting the development of AI PCs. Looking at the known AI PCs, it is not difficult to find that they all have a common point in the hardware level, that is, they all have the "NPU," a computing unit that was rarely seen in PCs before.Since 2014, the global television market has been trending towards saturation. For tech manufacturers, a wave of inward-competition stemming from hardware specifications, similar to the mobile phone industry, has swept from screen size to display quality, and finally towards the chip. As AI applications land on more super terminals such as mobile phones and PCs, large models have moved from the cloud to the terminal, and the television industry, once a super terminal, cannot ignore this catalyst that propels the industry forward.
Major tech manufacturers have successively implemented AI in television terminals, quietly initiating a transformation of AI televisions (AITV). As a leading player, Samsung is no exception. In addition to launching new TV products for 2024 on March 17th this year, it also proclaimed the slogan "AI for ALL," fully committing to AI. In 2024, Hisense Visual Technology has also entered the field of TV AI and released its self-developed Xinghai large model product. Hisense Visual Technology has also proposed three major technical foundations for TV AI, which are computing power, algorithms, and data. To better leverage the advantages of AI, Hisense has launched the Xinxin AI perception chip, and the image enhancement capabilities have also been improved. According to the official statement, this marks the full entry into the era of AI-calculated image quality.
03
Can AI technology truly stimulate new consumer demand?
The upgrade of terminal-side computing power brings higher performance requirements for AI large models on hardware, which helps to promote a recovery in demand for consumer electronic products. The main reason is that consumer electronic products such as AI mobile phones and AI personal computers (AIPC) will shorten the user's replacement cycle and accelerate the arrival of the replacement wave.
In terms of AI mobile phones, the quantity and quality of parameters largely determine the capabilities and performance of AI large models. The more parameters a large model contains, the higher the accuracy, and the more powerful the generative AI performance on mobile phones. Currently, Android flagship models such as the VIVO X100 series, OPPO Find X7 series, and Honor Magic6 series have successfully implemented local deployment of 7 billion LLMs. It is expected that the upper limit of local large model parameters will increase to 13 billion in 2024, and is expected to reach 17 billion in 2025.
Larger parameter-level models set higher requirements for mobile hardware configurations. AI computing power is the top priority for SOC upgrades. As the most important AI computing unit in mobile phones, SOC/AP largely determines the scale of AI large models deployed locally, which in turn determines the generative AI capabilities of mobile phones. Mobile SOCs have generally begun to integrate independent AI computing units such as APUs or NPUs, specifically responsible for handling heavy AI tasks. In addition, AI large models require more power to achieve high-frequency inference tasks, so terminal devices need to be equipped with larger-capacity batteries and higher-power chargers, thereby promoting the upgrade of power management chips. At the same time, when running large models, they need to reside in memory, and each processing of generative AI tasks will involve the transfer of massive amounts of data, which has higher requirements for the capacity and performance of mobile memory and flash memory. Under these circumstances, AI smart phones will stimulate a new wave of consumer demand.
In terms of AIPC, as the application ecosystem of AIPC becomes more mature, the user experience will be comprehensively improved, which will stimulate consumer demand for purchasing machines, and AIPC will drive the PC market into a new round of growth. Microsoft believes that the Copilot+ PC hardware configuration requires at least 16GB of memory (DDR5/LPDDR5), 256GB SSD/UFS, and the integrated performance of the NPU needs to reach more than 40 TOPS; Intel states that the entry-level standard for future AIPC will be upgraded to 32GB of memory, and 64GB PCs will begin to be shipped in 2025. In addition, AIPC needs to be equipped with high transmission rate memory to shorten the response time of large models, and the proportion of LPDDR with higher transmission rate is expected to further increase. Under these circumstances, the industry expects that the proportion of users replacing PCs within two years will double, reaching 20% or even higher.
In terms of AITV, as of this June, the sales volume of Samsung's new high-end AI TVs with a screen size of 75 inches (189 cm) or above has increased by more than 40% year-on-year. Enhanced artificial intelligence features are considered a key factor in the growing ultra-large high-end TV market. These features provide a differentiated viewing experience through AI-based image magnification and sound optimization. The use of AI energy-saving mode is also one of the popular features.
AI-empowered consumer electronics is a future trend. With AI-empowered consumer electronics, it will inevitably stimulate new consumer demand.
Leave a Reply