01/10 2025 386
At the Consumer Electronics Show (CES), the world's largest consumer electronics exhibition recently held in Las Vegas, USA, over 4,000 companies from 160 countries and regions participated, including more than 1,300 Chinese firms and 309 Fortune 500 companies such as NVIDIA, AMD, and Lenovo.
At this prestigious event, AMD, NVIDIA, and Intel all made significant strides in promoting "smart hardware" based on AI technology, with the majority of their new products centered around AI. It is evident that edge AI has emerged as a new trend, gradually making its way into the homes of ordinary people.
In recent years, nearly all chip hardware has begun to integrate AI functions, making AI capabilities indispensable for any device. Conversely, AI large models cannot exist without the support of hardware, as any new product must address three pivotal issues: its application scenario, market acceptance, and profitability.
AIPC, or Artificial Intelligence Personal Computer, is a type of personal computer that integrates AI technology. By incorporating hardware such as NPU, CPU, and GPU, it achieves high performance with low power consumption, fundamentally transforming, reshaping, and reconstructing the PC experience, thereby unleashing users' productivity and creativity. AIPC can be applied in various scenarios, including graphics vision, semantic understanding, and intelligent interaction, and it also drives the evolution of traditional PCs from mere tools to assistants, becoming individuals' "second brains" and AI companions.
The concept of AIPC was first introduced in September 2023, and it truly took off in 2024, making 2024 the inaugural year of AIPC. At this year's CES exhibition, major manufacturers took a significant step forward by launching new products.
On May 20, 2024, Microsoft announced the release of Windows 11 AIPC and declared the full integration of its AI assistant Copilot into the Windows system. Copilot leverages the GPT-4 model introduced by OpenAI. Simultaneously, Microsoft defined AIPC based on the "Copilot + PC" standard, emphasizing the need for a built-in NPU (Neural Processing Unit) with computing power exceeding 40 TOPS and memory exceeding 16GB to ensure that the computer can independently run AI computations without relying on network servers.
Currently, chip manufacturers such as Intel, AMD, Qualcomm, and NVIDIA, which are upstream in the AIPC industry chain, are continuously introducing high-performance processors for AIPC.
At CES, Intel unveiled the new Core Ultra processors (second generation), including the new Intel Core Ultra 200HX and 200H series mobile processors, the new Intel Core Ultra 200V series mobile processors based on the Intel vPro platform, the new Intel Core Ultra 200U series mobile processors, the expanded Intel Core Ultra 200S series desktop processors, and new Intel Core Ultra processors and Intel Core processors designed for edge computing. Intel claims that compared to the previous generation, the Core Ultra 200H series offers approximately 17% and 20% improvements in single-threaded and multi-threaded performance, respectively, a 22% enhancement in integrated graphics performance, and AI computing power reaching 99 TOPS.
AMD also announced the launch of the new Ryzen AIMax series processors to meet the demand for high-performance computing in high-end thin-and-light laptops. Concurrently, it introduced the new Ryzen AI300 series processors based on the "Zen 5" architecture, further enriching its product lineup. To continue the advantages of AMD's "Zen 4" architecture, AMD also launched the Ryzen 200 series processors for daily office use and expanded its commercial AIPC product line by integrating AMD PRO technology into the Ryzen AIMax, Ryzen AI300, and Ryzen AI200 series processors. Thanks to the integration of an NPU with up to 50 TOPS of AI processing power, the Ryzen AIMax series processors have become a potent driving force for the next generation of AI computers.
NVIDIA is not to be outdone. At CES, Huang Renxun unveiled the RTX 50 series graphics cards for PCs, continuing to provide robust power support for AIPC. Moreover, NVIDIA has directly entered the AIPC arena by introducing the small supercomputer Project Digits, enabling more developers to utilize large models on desktops. The new GeForce RTX 50 series adopts the NVIDIA Blackwell architecture, fifth-generation Tensor Cores, and fourth-generation RT Cores, with upgrades in the field of AI rendering, including neural network shaders, digital human technology, geometry, and lighting.
It is worth mentioning that NVIDIA's supercomputer Project Digits, equipped with the new GB10 superchip, is the world's smallest AI supercomputer capable of running 200B parameter models. MediaTek also participated in the design of GB10. The GB10 superchip in Project Digits uses the Blackwell GPU and is interconnected with a high-performance Grace CPU via NVLink-C2C inter-chip connectivity. This CPU adopts the ARM architecture and features 20 energy-efficient cores.
Last year, Qualcomm released the Snapdragon X Elite and Snapdragon X Plus processors and launched some PC products. However, in the overall Windows PC market, devices using the Snapdragon X series still account for less than 1.5%. In October last year, Qualcomm suddenly canceled its developer kits. Nonetheless, at CES, Qualcomm made a comeback by introducing the new system-on-chip Snapdragon X, targeting the mid-range PC market. This chip adopts a 4-nanometer process technology, offering long battery life and powerful performance, supporting the latest AI software, and being available in PCs priced as low as $600.
With the support of chip manufacturers, computer manufacturers such as Lenovo, HP, and Dell have increased their R&D and production investments in the AIPC field and launched related computer products. At CES, AIPC terminal manufacturers also announced new products.
Lenovo unveiled a series of new products, including the YOGA Air 14 Aura AI, ThinkPad X9, tablets and handheld gaming devices equipped with local AI, and the latest gaming laptops featuring NVIDIA RTX 5090 graphics cards. Dell introduced a new AIPC product portfolio specifically designed for personal and professional computing: Dell, Dell Pro, and Dell Pro Max, similar to Apple's naming scheme. ASUS also showcased a product known as the "lightest Copilot + PC".
Apart from the Windows ecosystem, Apple also launched its AI suite in June 2024, unveiling its first generative AI large model, Apple Intelligence, laying out a comprehensive strategy in the AIPC field.
Besides AIPC, smartphone manufacturers have also ignited a wave of AI competition. Nowadays, more and more smartphone manufacturers are integrating AI large models into their phones, making 2024 the inaugural year of AI smartphones.
Li Weimin, Deputy Director Analyst at the Strategic Development Research Institute of China Telecom Research Institute, pointed out that in terms of interaction logic, natural language processing capabilities have been continuously improved, equivalent to voice assistants possessing "high IQ" that can accurately discern users' intentions. Multimodal interaction combines voice, gestures, facial expressions, etc., making the operation interface more natural and rich. At the software ecosystem level, developer tools such as smart cameras and smart translation are also becoming intelligent, enhancing both the efficiency and quality of development and testing. Li emphasized that AI will undoubtedly serve as a "catalyst" in the smartphone market competition. Brands are now increasing their investments in AI research and development because those that can break through in this field are expected to reach the pinnacle.
At Apple's fall product launch event in 2024, Tim Cook demonstrated the Apple Intelligence feature. The new iPhone 16 series became Apple's first true AI smartphone. On October 22, Huawei officially released HarmonyOS NEXT, which deeply integrates AI with the OS, introducing a new HarmonyOS-native intelligence, Harmony Intelligence, and ushering in an OS experience tailored for the era of AI large models. Based on a powerful AI architecture, the Xiaoyi intelligent agent, equipped with the Pangu large model, is deeply integrated with the system's AI navigation bar and resides on the screen, becoming a readily available system-level intelligent agent.
At its October new product launch event, Honor introduced smartphones equipped with a new AI system. To demonstrate the power and intelligence of its AI features, Honor CEO Zhao Ming used the AI function to order 2,000 cups of Luckin Coffee for guests on-site. Additionally, Xiaomi launched "Super Xiao'ai," vivo unveiled "Lanxin Large Model," and OPPO released "AndesGPT".
By the end of 2024, rumors circulated in the market that Xiaomi Group was actively building its own GPU cluster of 10,000 cards and increasing its investment in AI large models. Xiaomi's large model team initially had access to 6,500 GPUs. This move underscores Xiaomi's determination and investment in the AI hardware field. Last May, Xiaomi also announced that its large language model, MiLM, had officially passed the large model registration. Xiaomi stated at the time that its large model would gradually be applied to Xiaomi's automobiles, smartphones, smart homes, and other products, enabling multi-device collaboration within and between scenarios through the combination of endpoints and clouds, empowering the "people-car-home ecosystem" strategy.
Apart from terminal manufacturers, MediaTek and Qualcomm's contributions in the AI field are also noteworthy. Their new products, such as the Dimensity 9400 and Snapdragon 8 Elite, provide robust technical support for the rapid development of AI smartphones.
The Dimensity 9400 integrates MediaTek's Dimensity Agentic AI Engine and the eighth-generation AI processor NPU 890. This combination not only significantly enhances the performance of traditional AI applications but also pushes them to a new height of highly intelligent AI applications featuring autonomous perception, reasoning with "brains," and collaborative action. This means that the Dimensity 9400 can fully empower edge AI, enabling rapid intelligent transformation of traditional applications across various fields such as text processing, image processing, and music creation.
The Snapdragon 8 Elite is equipped with a newly upgraded Hexagon NPU, offering innovative multimodal model processing capabilities that enable AI computing and computer vision to work synergistically, thereby significantly improving the execution efficiency of AI tasks. For instance, through LMM (Multimodal Model), the NPU can simultaneously process voice commands and image content, achieving faster response speeds. This efficient processing capability allows the Snapdragon 8 Elite to excel in various smart application scenarios.
Observing the development trends of various brands' developer conferences, the focus has shifted from simply optimizing the original functions of smartphones to centering on AI agents, leveraging localized edge models to provide users with more personalized services and bring new experiences to mid-to-high-end consumers. Deeply integrating AI capabilities with their own operating systems has become an important feature of flagship smartphones from major manufacturers and a crucial opportunity for them to overtake competitors in brand and market competition.
Smart hardware is a technological concept that follows smartphones, transforming traditional devices through the combination of software and hardware to endow them with intelligent functions. After being intelligentized, hardware gains connectivity capabilities, enabling the loading of Internet services and forming a typical "cloud + end" architecture, which provides added value such as big data.
On this basis, smart hardware has extended from wearable devices to fields such as smart TVs, smart homes, smart cars, healthcare, smart toys, and robots. Nowadays, with the support of AI, various new products such as MCUs, MPUs, and SOCs are gradually integrating AI functions, further fostering breakthroughs in smart hardware.
In 2024, ByteDance released its first AI agent earphone, Ola Friend, which connects to its Doubao large model and integrates with the Doubao APP. Thus, this AI smart hardware represents ByteDance's exploration and experimentation in AI scenarios, providing users with more direct assistance in various aspects of their lives through the Doubao large model.
At CES 2025, Samsung introduced the industry's first monitor that integrates AI into the device end, while LG unveiled new smart fridges and microwaves that utilize AI for intelligent control and energy management, automatically adjusting temperature, operation modes, etc., based on users' usage habits.
Recently, Google and Synaptics also announced their collaboration in the field of Edge AI for the IoT, aiming to determine the optimal solution for multimodal processing in context-aware computing. This collaboration will integrate Google's MLIR-compliant ML cores and open-source software and tools onto Synaptics' Astra hardware, accelerating the development of IoT AI devices and supporting the processing of various modalities such as vision, images, speech, and sound, providing seamless interaction contexts for applications in wearable devices, home appliances, entertainment, embedded hubs, monitoring, and control in consumer, automotive, enterprise, and industrial systems.
The "AI+" trend is not only a microcosm of technological development but also reflects how technology plays a pivotal role in enhancing people's quality of life and driving social progress. Through the integration of smart hardware and AI technology, future life will become more convenient, efficient, and intelligent.
The integration of AI has injected new vitality into the consumer markets for PCs, smartphones, and smart hardware.
In the PC market, IDC predicts that the Chinese PC market will end its negative growth due to AIPC and maintain a steady growth trend over the next five years. The total size of the PC market will increase from 39 million units in 2023 to over 50 million units in 2027, representing a growth rate of nearly 28%. The proportion of new AIPC installations in the Chinese PC market will rapidly increase in the coming years, reaching 85% by 2027 and becoming the mainstream in the PC market.
In the realm of smartphones, industry experts anticipate that the integration of AI applications on mobile devices by 2025 will propel a fresh wave of replacement cycles, heralding a steady recovery for the global smartphone market. According to a research report from China Merchants Bank International, this integration is set to intensify the replacement cycle, forecasting a 3% year-on-year increase in global smartphone shipments, reaching 1.25 billion units.
Fan Zhiyuan, Chief Analyst of Electronics at Guojin Securities, holds that by 2025, AI applications will sustain the upward momentum of the semiconductor cycle. Innovations spurred by generative AI are poised to dominate the AI landscape, predominantly presented as "AI+X," encompassing a wide array of products and services such as AI smartphones, AIPC, AI glasses, AI headphones, AI speakers, AI players, AI education, AI companionship, robots, and autonomous driving. The evolution and innovation of these terminal demands will fuel the demand for chips, thereby fueling the continuous expansion of the entire semiconductor market.
Amidst this new development paradigm, the optimal application of AI has emerged as a focal point. Leveraging their resources and technological prowess, numerous technology giants are actively integrating AI into their core businesses, reimagining traditional products, and enhancing their market competitiveness. Simultaneously, they are deploying cutting-edge technologies to broaden AI application scenarios. Furthermore, a multitude of startups are delving into the potential of AI, achieving significant advancements and transformations in functionality and user experience through innovative product development or pioneering downstream hardware innovations.