Core Viewpoint - The launch of the scaleX 10,000-card AI supernode by Zhongke Shuguang marks a significant milestone in China's AI computing power history, entering the era of 10,000-card supernodes [1][3]. Group 1: Development of AI Computing Power - The establishment of the scaleX 10,000-card supernode represents a new answer to the development path of China's AI computing infrastructure [3]. - Three years ago, China's AI computing power system heavily relied on NVIDIA for GPU acceleration, NVLink technology, and CUDA software, creating a dependency on a single supplier [4]. - The turning point came with export restrictions on NVIDIA chips, prompting domestic manufacturers to explore alternative computing power systems [4]. Group 2: Competitive Landscape - Major players like Huawei, Inspur, and Alibaba are entering the AI supernode market, each adopting different technological routes [5]. - Huawei has taken a "fully self-developed" approach, while Inspur and Alibaba focus on "open architecture" to build a domestic AI computing foundation [6]. - The scaleX 10,000-card supernode consists of 16 scaleX640 supernodes, totaling 10,240 AI accelerator cards and exceeding 5 EFlops in computing power [7]. Group 3: Technological Innovations - The scaleX640 supernode features a self-developed scaleFabric high-speed network with a bandwidth of 400 Gb/s and an end-to-end latency of less than 1 microsecond [7]. - The system supports multiple brands of accelerator cards, indicating a shift towards a diversified computing power ecosystem in China [7]. Group 4: Industry Trends - The trend of "de-NVIDIA" is driven by the need for computing power security and independent innovation in China, especially following U.S. export restrictions on high-performance GPUs [8]. - The domestic AI industry is not merely replicating NVIDIA but aims to establish a complete, replaceable computing power ecosystem [8]. - The development paths of closed-stack integration, represented by Huawei, and open collaboration, represented by Shuguang, Inspur, and Alibaba, are emerging as two significant trends in the industry [8]. Group 5: Application and Impact - Various products have already been deployed, with Huawei's CM384 and Inspur's SD200 being used in operational data centers [9]. - The open architecture approach has facilitated the large-scale application of domestic chips, moving away from reliance on NVIDIA's ecosystem [9]. - The year 2025 is seen as a turning point for China's AI computing power system, emphasizing the importance of both performance and collaborative ecosystems [11].

一、宏观迷雾：为什么"坏数据"成了"救命稻草"？ | 首页 | 昼念板块 财报日历 | 新股中心 | 机构造综 | 财经日历 | | | | | | | | | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 道琼斯指数 | | | | 纳斯达克综合指数 | | | | 标准500指数 | | | | | | 48114.26 | | | | 23111:46 | | | | 6800.26 | | | | | | 305 30 -D 03 | | | | *24.02 +0.23 | | | | 16:25 -0 24 | | | | | | 领涨榜 | 盘中 ·· | | 更多 > | 领跌榜 | 国中 · | | 更多 > | 热力图 行业 个股 | | | | 更多 > | | 代码 | នធ | 新颜描 : | 履新价 | 代码 | 名称 | 源联盟 : | 量新价 | | | | | | | VMAR | Vision Marine | +156.77% | 0.6760 | TTS ...

Core Viewpoint - The domestic AI chip market in China is undergoing a significant transformation, with a focus on "de-NVIDIA" efforts led by major tech companies like Alibaba and Baidu, aiming to challenge NVIDIA's dominance in the AI chip sector [1][3]. Group 1: Market Dynamics - Chinese tech giants are actively promoting the development of self-researched AI chips, with Alibaba and Baidu announcing that their core AI models will partially utilize self-developed chips [1][3]. - Since late August, the stock prices of Baidu and Alibaba have surged by approximately 50% [1]. - The geopolitical tensions and concerns over the stability and security of the AI supply chain are driving the "de-NVIDIA" movement in China [3][5]. Group 2: NVIDIA's Challenges - NVIDIA faced a significant negative impact due to export restrictions on its H20 chip, leading to a stock impairment of about $4.5 billion in Q1 [5]. - Revenue from mainland China for NVIDIA dropped to $2.77 billion in Q2 of FY2026, a nearly 50% decline, reducing its market share from 85% to 70% in China [5][11]. Group 3: Rise of Domestic Chips - Domestic custom AI chips are rapidly emerging, with products like Alibaba's PPU chip and Huawei's Ascend series showing performance that rivals or exceeds NVIDIA's offerings [7][9]. - The PPU chip's single-card cost is approximately 40% lower than the imported H20 chip, highlighting the cost advantage of domestic solutions [7]. - IDC forecasts that by 2024, domestic AI chip brands will significantly increase their market share to 30% [11][13]. Group 4: Industry Evolution - The shift towards customized AI chips mirrors the evolution of smartphone chips from generic to specialized designs, driven by the need for better performance and cost efficiency [16][19]. - The transition from general-purpose GPUs to customized chips is essential for meeting the specific demands of AI inference tasks, which require lower power consumption and reduced latency [20][21]. - The development of domestic chip design and supply chains is enabling Chinese companies to enhance their competitiveness in the global market [23][24].

Core Viewpoint - Nvidia has become the first publicly traded company to surpass a market capitalization of $4 trillion, achieving this milestone in just over two years since reaching $1 trillion, highlighting the rapid growth in the AI sector and the dominance of computing power [4][5]. Group 1: Nvidia's Market Position - Nvidia's market value growth is one of the fastest in Wall Street history, emphasizing the importance of computing power in the AI era [5]. - Despite Nvidia's success, competition is increasing as major cloud service providers like Google, Amazon, and Microsoft are developing their own ASIC chips while using Nvidia's GPUs [5][21]. - Nvidia's GPUs currently dominate over 80% of the AI server market, while ASICs account for only 8% to 11% [21]. Group 2: ASIC Market Dynamics - The growth of ASICs is a response to changing industry demands rather than a cause, with ASICs being tailored for specific applications in AI [12][13]. - As AI model development progresses, the demand for ASICs is expected to rise, complementing the existing GPU market rather than replacing it [19][20]. - The rapid growth of ASICs indicates a significant maturation of application-side demand in North America, driven by the explosion of AI token usage [19]. Group 3: Competitive Strategies - Nvidia's recent introduction of NVLink Fusion allows for the integration of Nvidia GPUs with third-party CPUs or custom AI accelerators, breaking down previous hardware ecosystem barriers [23][25]. - This semi-open NVLink Fusion strategy is seen as a defensive move against ASIC competitors while maintaining Nvidia's ecosystem advantages [25][28]. - The emergence of UALink, initiated by major tech companies, aims for higher openness compared to Nvidia's NVLink, but is still in the early stages of development [27][28].

Core Viewpoint - Major cloud service providers and Nvidia's clients are beginning a long "divorce" process, focusing on developing their own ASIC chips to reduce dependence on Nvidia's expensive hardware and software ecosystem [1][2]. Group 1: Market Trends - The procurement of Application-Specific Integrated Circuits (ASICs) is expected to grow at a compound annual growth rate (CAGR) of 50%, primarily driven by companies like Microsoft, Google, and Amazon AWS [1]. - Nvidia's hardware, particularly the Blackwell architecture B200 GPU, is widely used in data centers, but its high cost (ranging from $70,000 to $80,000 per chip) is prompting clients to seek alternatives [1]. Group 2: Client Strategies - Core cloud computing clients of Nvidia are increasing their orders for ASIC hardware while still purchasing Nvidia products, indicating a gradual shift towards hardware autonomy [2]. - Companies like Amazon and Google are heavily investing in self-developed chips, with Amazon reportedly running about 50% of its new servers on its AWS Graviton Arm processor family [3]. Group 3: Industry Dynamics - Nvidia is forming partnerships with various ASIC manufacturers through its NVLink Fusion program, allowing seamless collaboration between Nvidia hardware and third-party ASIC servers [3]. - TSMC, as a major foundry for both Nvidia's hardware and the ASIC chips of large cloud clients, is positioned to benefit significantly from this trend [3].