Core Insights - The emergence of SuperPods as a new AI computing infrastructure has become a focal point in the industry since 2025, with Huawei's Ascend 384 SuperPod leading the way in performance metrics compared to foreign competitors [1][3] - The demand for computing power is far from being met, with token consumption expected to exceed trillions daily in China, highlighting the inadequacy of simply stacking servers to address the computing gap [3][4] Group 1: SuperPod Characteristics - SuperPods are not merely about stacking chips; they represent a fundamental restructuring of traditional computing architectures, enabling equal interconnectivity among CPUs, NPUs, and memory units [4][6] - Key features of a true SuperPod include high bandwidth to eliminate communication delays, low latency, and the ability to form a logically unified system through unified memory addressing [6][7] Group 2: Efficiency and Performance - SuperPods can significantly enhance computing efficiency, with potential model utilization rates increasing from 30% to 45%, effectively a 50% improvement, which can help mitigate the limitations of chip manufacturing processes [7][8] - The architecture of SuperPods differs from traditional systems, as Huawei employs optical communication technology, allowing for greater scalability and interconnectivity compared to NVIDIA's copper-based systems [8][9] Group 3: Innovation and Ecosystem - Huawei's systematic innovation in chip design, optical components, and foundational protocols has positioned it uniquely in the market, leveraging over 20 years of experience in optical technology [9][12] - The company is also developing general computing SuperPods, with the TaiShan 950 SuperPod set to launch in Q1 2026, aimed at replacing various server applications [11][12] Group 4: Software and Community Engagement - The success of SuperPods relies not only on hardware but also on a robust software ecosystem, including open-source initiatives like CANN and openEuler, which are crucial for fostering industry collaboration [14] - Huawei has engaged a large developer community, with 3.8 million registered developers for Kunpeng and nearly 4 million for Ascend, emphasizing the importance of open-source collaboration in the AI era [14]

Core Viewpoint - The emergence of SuperPod as a new AI computing infrastructure has become a focal point in the industry since 2025, with Huawei's Ascend 384 SuperPod leading the way in performance metrics compared to foreign competitors [1][3]. Group 1: SuperPod Concept and Advantages - SuperPod is not merely about stacking chips; it represents a fundamental restructuring of traditional computing architecture, enhancing communication efficiency among CPU, NPU, and memory units [4][6]. - The key advantages of SuperPod over traditional clusters include significantly improved computational efficiency, with potential model computing utilization rates increasing from 30% to 45%, equating to a 50% performance boost [7][8]. Group 2: Technical Challenges and Innovations - Building a true SuperPod is complex; Huawei's Ascend 384 SuperPod consists of 12 computing cabinets and 4 bus cabinets, while NVIDIA's NVL72 system is confined to a single cabinet due to architectural differences [8]. - Huawei employs optical communication technology for interconnection, allowing for greater scalability beyond single cabinet limitations, while traditional systems face constraints with electrical signal transmission [8][9]. Group 3: Systematic Innovation and Ecosystem Development - Huawei's systematic innovation includes proprietary chip development, optical device capabilities, and foundational protocols, enabling the creation of SuperPods that leverage full optical interconnectivity [9][12]. - The company is also developing general computing SuperPods, such as the TaiShan 950, which aims to replace various server applications by 2026 [9][11]. - A robust software ecosystem, including open-source initiatives like CANN and openEuler, is essential for the operation of SuperPods, with a focus on collaborative development within the industry [14].

Economic News - The Ministry of Finance, General Administration of Customs, and State Taxation Administration announced a "zero tariff" policy for imported goods purchased by residents of Hainan Free Trade Port, allowing an annual exemption limit of 10,000 yuan per person [1] - The Shanghai Futures Exchange will adjust the price fluctuation limits and trading margin ratios for gold and silver futures starting February 9, 2026, with gold futures' fluctuation limit increasing from 16% to 17% [2] - The Ministry of Industry and Information Technology and other departments released a plan for the high-quality development of traditional Chinese medicine, aiming to cultivate 60 high-standard raw material production bases by 2030 [3] - The Ministry of Commerce reported a 7.4% year-on-year growth in service trade in 2025, with knowledge-intensive service trade growing by 6.6% [4] - The National Internet Information Office and 11 other departments issued opinions to enhance digital services for foreign personnel entering China, aiming for a more interconnected digital service system by 2030 [5] Key Company Information - Meituan announced the acquisition of Dingdong Maicai's China business for approximately $717 million (about 5 billion yuan) [6] - Guoxuan High-Tech plans to raise 5 billion yuan through a private placement to expand its new energy battery production capacity [7] - Nanjing Mining Group intends to acquire a 10% stake in Eagle Valley Gold through a cash investment of $30 million [8] - Muyuan Foods reported a 2.73% year-on-year increase in the sale of live pigs in January 2026, with sales revenue declining by 11.93% [9] - Baidu announced a new stock repurchase plan of up to $5 billion, effective until December 31, 2028, to enhance long-term shareholder value [10] Industry Insights - The rubidium and cesium market is expected to enter a new structural expansion cycle driven by the penetration of perovskite solar cells and the development of space photovoltaics [11] - Perovskite solar cells are projected to have a market penetration rate of 30% by 2030, significantly increasing from 1.3% in 2025 [12] - The flexible structure of perovskite solar cells allows for applications in various fields, including building-integrated photovoltaics (BIPV), wearable devices, and vehicle power generation [13] - The global BIPV market is expected to grow from $16.66 billion to $47.02 billion between 2026 and 2031, with a CAGR of 23.06% [14] - The demand for rubidium is projected to grow at a CAGR of 115% from 2025 to 2030, driven by the increasing production of perovskite solar cells [15]

Core Viewpoint - Starting from 2025, supernodes will become a significant technological innovation direction in the AI computing network, with increasing competition among AI chip manufacturers in both chip performance and Scale up network [1][5]. Group 1: Supernode Development - Nvidia has launched mature supernode solutions, with plans to release GH200 NVL72, GB200/GB300 NVL72, and VR200 NVL72 from 2024 to 2026 [1][3]. - The Blackwell architecture standardizes Scale up with GB200 NVL72 stabilizing the scale at 72 GPUs per cabinet, consisting of 18 Compute Trays and 9 Switch Trays [2]. - The Rubin architecture will enhance bandwidth, with the NVLink 6 Switch increasing single GPU interconnect bandwidth to 3.6 TB/s, up from 1.8 TB/s [2]. Group 2: Nvidia's Competitive Advantage - Nvidia maintains a leading position in the supernode market, with a projected shipment of approximately 2,800 units of GB200/300 NVL72 by 2025 [3]. - Future plans include the introduction of Vera Rubin NVL144 and Rubin Ultra NVL576, expanding interconnected GPUs from 72 to 576 [3]. - Innovations such as NVLink and NVLink Switch are crucial for achieving high bandwidth and low latency in AI training clusters, with NVLink 5 Switch supporting a total bandwidth of 130 TB/s for 72 GPUs [4]. Group 3: Industry Landscape and Investment Strategy - The global supernode competition landscape is still being established, with Nvidia currently in a leading position [6]. - The report suggests monitoring Nvidia's supernode supply chain, including components like PCB backplanes, high-speed copper cables, optical modules, and cooling systems [6]. - Chinese manufacturers are actively participating in the supernode and Scale up network sectors, with potential for domestic firms to gain a competitive edge [6].

Investment Rating - The report maintains a "Positive" outlook on the communication industry [2] Core Insights - The evolution of large language model (LLM) parameters from hundreds of billions to trillions and even hundreds of trillions necessitates tensor parallelism (TP) across servers, making the development of high-bandwidth, low-latency Scale up networks a mainstream technical path in the industry [4][18] - NVIDIA is positioned as a leader in the supernode space, with plans to launch multiple generations of supernodes from 2024 to 2026, including GH200 NVL72, GB200/GB300 NVL72, and VR200 NVL72 [5][43] - The advantages of NVIDIA's supernodes are built on NVLink and NVLink Switch technologies, which support high bandwidth and low latency data transmission essential for AI training clusters [6][86] Summary by Sections 1. High Bandwidth and Low Latency Requirements - The training of LLMs requires extremely high bandwidth and low latency, driving the innovation of supernodes as a key direction in AI computing networks [18] - The need for cross-server tensor parallelism (TP) and expert parallelism (EP) has led to the establishment of Scale up networks [8] 2. NVIDIA's Leading Advantage - NVIDIA's supernode solutions are based on NVLink and NVLink Switch, which have evolved from point-to-point connections to full interconnect communication [33] - The sixth generation of NVLink and NVLink Switch supports GPU-to-GPU communication bandwidth of 3.6TB/s, with total aggregated bandwidth of 260TB/s in the VR NVL72 system [33][75] 3. Supernode Specifications - The GB200 NVL72 supernode features 180 PFLOPS of TF32 Tensor Core computing power, 13.8TB of memory, and a memory bandwidth of 576TB/s, with a total exchange capacity of 129.6TB/s [47][48] - The VR200 NVL72 supernode, set to be released in 2026, will double the total exchange capacity to 259.2TB/s compared to the GB200 NVL72 [70][75] 4. Investment Strategy - Starting from 2025, supernodes will become a significant innovation direction in AI computing networks, with various global manufacturers entering the competition [9] - NVIDIA currently holds a leading position, and attention should be paid to its supernode supply chain, including PCB backplanes, high-speed copper cables, optical modules, and cooling systems [9]

Summary of Conference Call Records Companies and Industries Involved - **Companies**: Haiguang Information, Lianqi Technology, Wangsu Technology - **Industries**: AI computing, CDN (Content Delivery Network), semiconductor technology Key Points and Arguments Haiguang Information - Haiguang Information's market capitalization increased by over 90 billion RMB, leading the A-share market in January 2026 [2] - The company’s Deep Computing 3 has entered mass production, supporting FP8/FP4 precision, while Deep Computing 4 is expected to double performance, potentially becoming the strongest AI chip in China [3][7] - The estimated valuation for Haiguang's CPU is 900 billion RMB and for its GPU is 1.3 trillion RMB [3][7] - The company is projected to reach a market capitalization of over 2 trillion RMB by 2028, with a target of 1.2 trillion RMB for 2026 [8] Lianqi Technology - Lianqi Technology benefits from the growth in AI inference and supernode industries, particularly in memory interconnect chips, PCIe Retimer Switch, and CXL chips [1][2] - The company has made significant progress in the CXL field, with its products expected to be adopted by Google's next-generation TPU, creating a substantial incremental market [10] - Lianqi's revenue breakdown includes 90% from memory interconnect, 5% from PCIe CXL, and 5% from CPU and server-related products [10] Wangsu Technology - Wangsu Technology is the largest third-party neutral CDN company in China, with CDN business accounting for 60-70% of its revenue [11] - The company is benefiting from a near doubling of CDN prices in North America due to Google Cloud's price increase, indicating a reversal in the CDN and cloud computing price war [1][2][12] - Wangsu is expected to achieve a net profit of 1 billion RMB in 2026, with significant profit elasticity due to price increases, suggesting over 50% growth potential in its valuation [12] Capital Expenditure Trends - North America's top five CSPs are projected to have capital expenditures nearing 700 billion USD in 2026, a 50% increase from 400 billion USD in 2025, driven by Meta and Microsoft's unexpected capital spending [4] - Domestic internet capital expenditure in China is expected to reach 570-600 billion RMB in 2026, with growth anticipated to surpass that of overseas markets by 2027 due to advancements in self-developed chips and easing of restrictions [4] AI Inference Demand - The emergence of applications like MudBot is driving exponential growth in data and computing power consumption, shifting traffic from human-driven to robot-driven, enabling 24/7 usage [5] Supply-Side Technological Advances - Future server architectures are expected to adopt supernode technology, which will enhance cluster efficiency through memory pooling and high-speed interconnects [6] Other Notable Companies - Additional companies to watch include DingTong Technology, Zhongke Shuguang, Shuguang Shuchuang, Feirongda, Yingweike, and application vendors like Shuiyou Co. and Keda Xunfei, all of which show promising development prospects [13]

Investment Rating - The industry investment rating is maintained as "Positive" [2] Core Insights - The report emphasizes the transition to the "SuperNode" era in computing infrastructure, driven by the increasing computational demands of AI models, which are evolving from pre-training to multi-faceted scaling [14][17] - The SuperNode architecture relies on enhanced interconnectivity, integrating multiple computing chips into a logical "super GPU/ASIC" to overcome the limitations of traditional single-server setups [39] - The report identifies three main investment themes: "Network End + AIDC + Computing End" and four key sectors: "Optical + Liquid Cooling + Power Supply + Chips" [7] Summary by Sections 1. Transition to SuperNode Era - The evolution of AI models necessitates a shift in computational infrastructure, with demands for higher performance and efficiency [14] - The report highlights the rapid growth in model parameters and training data, indicating a need for advanced computational clusters [14] 2. SuperNode Architecture - SuperNode clusters are defined as large-scale GPU/ASIC systems that expand from thousands to millions of computational units, utilizing both Scale Up and Scale Out strategies [17] - The architecture includes components such as computing nodes, switch trays, and power supply units, which are essential for the operation of SuperNode systems [80] 3. Upgrades in Key Components - The report notes that the increasing penetration of SuperNode servers will drive demand for high-power supplies, liquid cooling systems, and optical communication technologies [5][80] - Specific companies are highlighted as beneficiaries in the sectors of optical modules, liquid cooling, server power supplies, and exchange chips [7] 4. Domestic SuperNode Development - The report discusses the progress of domestic SuperNode solutions, particularly Huawei's Atlas 900 A3 and Atlas 960, which are designed to enhance the performance of domestic AI clusters [6][112] - The performance of domestic chips is shown to improve significantly when deployed in SuperNode configurations, effectively compensating for individual chip limitations [112]

Core Insights - The company, Huaqin Technology, reported a total R&D expenditure of 4.62 billion RMB for the first three quarters of 2025, representing a year-on-year increase of 23.7%, with an expectation to exceed 6 billion RMB for the entire year [2] Group 1 - The company emphasizes the importance of collaborative win-win strategies in its operations [2] - During the CES exhibition, Huaqin Technology showcased its core technological breakthroughs in consumer electronics, automotive electronics, PC, and AIoT, highlighting its R&D innovation capabilities and ecosystem collaboration [2] - The company's R&D investment focuses on a product layout of 3+N+3 and supports business growth needs, while also allocating resources for forward-looking research in areas such as acoustics, optics, thermodynamics, and radio frequency [2] Group 2 - Huaqin Technology has established an X-lab R&D laboratory to explore new technologies and product directions, including AI edge computing, super nodes, and automotive electronics [2] - The company maintains close collaboration with upstream and downstream partners in the industry chain to jointly explore smart hardware solutions, such as the high-level autonomous driving collaboration with NVIDIA on the Drive Thor platform [2]

Core Insights - The domestic AI chip industry is experiencing a significant transformation in 2025, driven by the launch of DeepSeek, which has accelerated the demand for AI computing power and domestic alternatives [2][3] - Major domestic companies like Cambrian, Moore Threads, and Muxi are making headlines with their stock performances and upcoming IPOs, indicating a capital market boom for domestic AI chips [2][5] - The market is witnessing a shift in focus from training to inference, with a projected market size for AI inference chips in China expected to grow from 162.6 billion yuan in 2024 to 310.6 billion yuan in 2025 [3][4] Trend 1: Domestic AI Chip Opportunities - The launch of DeepSeek has ignited enthusiasm for domestic AI chips, particularly in the inference sector, as companies like Huawei and Cambrian quickly adapt to support this new model [2][3] - The inference chip market is expected to see explosive growth, with significant contributions from companies focusing on inference applications, such as Huawei, Cambrian, and Muxi [3][4] Trend 2: IPO Surge of Domestic AI Chip Manufacturers - 2025 marks a pivotal year for domestic AI chip manufacturers, with several companies successfully listing on the stock market, including Moore Threads and Muxi, which saw substantial initial stock price increases [5][6] - Despite the IPO successes, the market share of domestic chip manufacturers remains low, with major players like Nvidia and AMD dominating the market [6][7] Trend 3: Nvidia's Market Dynamics - Nvidia has faced challenges in the Chinese market, including export restrictions and security concerns, which have opened opportunities for domestic chip manufacturers [9][10] - The approval of Nvidia's H200 chip for sale in China could impact the domestic market, as it offers competitive performance, although concerns about dependency on foreign technology persist [10][11] Trend 4: Advanced Process Limitations - Domestic AI chip manufacturers are constrained by limitations in advanced process technologies, with most using 7nm or 14nm processes compared to Nvidia's 4nm technology [12][13] - Companies are exploring alternative solutions, such as switching to domestic supply chains and developing "super nodes" to enhance performance despite process limitations [13][14] Outlook: Future Opportunities for Domestic AI Chips - The domestic AI chip market is expected to exceed 300 billion yuan by 2026, driven by the rapid development of intelligent computing centers and increasing AI demands from internet companies [15][16] - The industry is anticipated to split into two main directions: self-developed ASICs by CSPs and local suppliers, with a focus on lower-spec AI inference chips presenting significant growth opportunities [16]

Summary of Key Points from the Conference Call Industry Overview - The conference discusses the liquid cooling industry, focusing on advancements in technology and market opportunities, particularly in relation to the LoongArch N8 72 model and its implications for data centers and cooling solutions [1][2][3]. Core Insights and Arguments - **LoongArch N8 72 Design**: The LoongArch N8 72 features a fully liquid-cooled and modular design, eliminating fans and simplifying integration. The power consumption has increased to 2,300 watts while maintaining a heat density of 100 watts per square centimeter, which can be managed by single-phase cold plates [1][2][4]. - **Microchannel Technology**: New technologies like Microchannel Liquid Plates (MLCP) aim to enhance cooling efficiency by reducing thermal resistance and increasing heat exchange area. However, challenges such as yield control and leak protection hinder large-scale deployment in the short term [1][5]. - **Increased GPU Density**: The LoongArch Ultra and its associated chassis will adopt a vertical plug-in design, increasing GPU density from 72 to 576, necessitating adjustments in energy solutions to accommodate higher complexity and density [1][6]. - **International Liquid Cooling Solutions**: Overseas ASIC chip cooling solutions predominantly utilize liquid cooling, with companies like Meta and AWS implementing Advanced Liquid Cooling (ALC) solutions. Google’s TensorFlow 10 CPU supports 2 megawatts of power across eight cabinets [1][8][9]. - **Domestic Market Opportunities**: The domestic liquid cooling market is expected to see significant growth by 2026, driven by the need for high-performance computing solutions and the shortage of high-end chips, which pushes for the development of supernodes [1][10][11]. Additional Important Insights - **System Integration Needs**: Chinese liquid cooling manufacturers must provide comprehensive, one-stop solutions that integrate data centers, cabinets, and internal nodes to enhance competitiveness in overseas markets [1][3][13]. - **Challenges in High-Density Designs**: As physical density increases, challenges such as maintaining flow uniformity and ensuring welding quality become critical. Advanced welding techniques and optimized cooling system designs are necessary to address these issues [1][7][18]. - **Pump Technology Transition**: The shift from mechanical pumps to magnetic pumps in cooling systems is noted, as magnetic pumps eliminate leakage risks and improve reliability, although the performance of electronic pumps remains to be fully evaluated [1][20]. - **Material Considerations**: The selection of materials for cooling systems must consider aging, compatibility, and the effects of long-term exposure to liquids, ensuring reliability and preventing issues like deformation and blockages [1][21]. This summary encapsulates the key points discussed in the conference call, highlighting the advancements and challenges within the liquid cooling industry and the implications for future developments.