Investment Rating - The report maintains a positive outlook on the computer industry, rating it as "Overweight" [5][6]. Core Insights - The report highlights the launch of the orthogonal architecture single-cabinet super node by Alibaba Cloud and ZTE, which features a high density of 128 AI chips, marking a significant advancement in AI server technology [5][7]. - OpenAI is transforming from a large model provider to an interactive ecosystem creator, with significant user growth and the introduction of new tools like AgentKit and Codex [29][30]. - Key company updates include Hikvision's approval of a mid-term dividend plan and the expansion of AI products, as well as the acquisition of Kunyu Lancheng by Suochen Technology to enhance its satellite internet capabilities [42][50]. Summary by Sections 1. Orthogonal Architecture Single-Cabinet Super Node - On September 24, 2025, Alibaba Cloud launched the Pangu 128 super node AI server, which is the highest density product among domestic super nodes [7][24]. - The orthogonal architecture reduces transmission paths and enhances interconnect efficiency compared to traditional designs [15][18]. - This architecture is expected to become a new paradigm for domestic super nodes, especially as liquid cooling infrastructure improves [28]. 2. OpenAI's Ecosystem Development - OpenAI's developer conference introduced several new products, including AgentKit and Codex, which aim to reshape developer habits and create a new AI operating system [29][32]. - The user base for ChatGPT has grown significantly, with weekly users increasing from 100 million in January 2023 to over 800 million by August 2025 [30]. - The Apps SDK is positioned to become a new entry point for AI applications, allowing seamless integration of various software services within ChatGPT [32]. 3. Key Company Updates Hikvision - Hikvision approved a mid-term dividend plan, distributing 4 CNY per share, totaling approximately 3.6 billion CNY, marking its first mid-term dividend since its IPO [42][48]. - The company is expanding its AI applications across various sectors, including industrial manufacturing and urban governance [44][48]. Suochen Technology - Suochen Technology announced the acquisition of 55% of Kunyu Lancheng for 74.25 million CNY, enhancing its capabilities in satellite internet and aerospace applications [50][52]. - The acquisition aims to bolster the company's physical AI integration and expand its application scenarios [52][56]. Jizhi Jia - Jizhi Jia is positioned as a leader in the AMR (Autonomous Mobile Robot) sector, with a focus on warehouse fulfillment solutions [58]. - The company is expected to benefit from rising labor costs and increasing supply chain efficiency demands, with projected revenue growth of 33% to 36% from 2025 to 2027 [61].

Core Viewpoint - The article provides a detailed comparison of Alibaba's super node with NVIDIA's NVL72 and Huawei's CM384, focusing on GPU count, interconnect technology, power consumption, and ecosystem compatibility. Group 1: GPU Count - Alibaba's super node, known as "Panjun," utilizes a configuration of 128 GPUs, with each of the 16 computing nodes containing 4 self-developed GPUs, totaling 16 x 4 x 2 = 128 GPUs [4] - In contrast, Huawei's CM384 includes 384 Ascend 910C chips, while NVIDIA's NVL72 consists of 72 GPUs [7] Group 2: Interconnect Technology - NVIDIA's NVL72 employs a cable tray interconnect method using NVLink proprietary protocol [8] - Huawei's CM384 also uses cable connections between multiple racks [10] - Alibaba's super node features an orthogonal interconnect without a backplane, allowing for direct connections between computing and switch nodes, reducing signal transmission loss [12][14] Group 3: Power and Optical Connections - NVIDIA's NVL72 uses copper for scale-up connections, while Huawei's CM384 employs optical interconnects, leading to higher costs and power consumption [15] - Alibaba's super node uses electrical interconnects for internal scale-up, with some connections made via PCB and copper cables, while optical interconnects are used between two ALink switches [18][19] Group 4: Parameter Comparison - Key performance metrics show that NVIDIA's GB200 NVL72 has a BF16 dense TFLOPS of 2,500, while Huawei's CM384 has 780, indicating a significant performance gap [21] - The HBM capacity for NVIDIA's GB200 is 192 GB compared to Huawei's 128 GB, and the scale-up bandwidth for NVIDIA is 7,200 Gb/s while Huawei's is 2,800 Gb/s [21] Group 5: Ecosystem Compatibility - Alibaba claims compatibility with multiple GPU/ASICs, provided they support the ALink protocol, which may pose challenges as major manufacturers are reluctant to adopt proprietary protocols [23] - Alibaba's GPUs are compatible with CUDA, providing a competitive advantage in the current market [24] Group 6: Supply Chain Insights - In the AI and general server integration market, Inspur holds a 33%-35% market share, while Huawei's share is 23% [33] - For liquid cooling, Haikang and Invec are key players, each holding 30%-40% of the market [35] - In the PCB sector, the number of layers has increased to 24-30, with low-loss materials making up over 60% of the composition, significantly increasing the value of single-card PCBs [36]