Core Insights - The report titled "2025 Domestic AI Chip Software Ecosystem White Paper" highlights the competitive landscape of domestic AI chips, emphasizing the shift in user focus from hardware performance to the maturity, compatibility, and usability of the software ecosystem [1][10] - The software ecosystem is structured into four layers: foundational support layer, core tools layer, framework adaptation layer, and management monitoring layer, which collectively determine the value realization and commercialization of AI chips [1][14] - Domestic AI chips are categorized into specialized acceleration chips, general-purpose computing chips, and graphics computing chips, with notable manufacturers focusing on different aspects of the ecosystem [1][10] Group 1: Background and Significance - The report outlines the context of increasing international technological competition, which has led to significant advancements in domestic AI chip technology and market expansion [10][11] - The primary goal of the white paper is to systematically assess the development status of the domestic AI chip software ecosystem, providing a technical reference for industry, academia, and government [12][13] Group 2: AI Chip Software Ecosystem Composition - The AI chip software ecosystem serves as a technical hub connecting hardware capabilities with upper-layer applications, structured into four main modules: foundational support, core tools, framework adaptation, and management monitoring [14][20] - Each module plays a critical role in the training and inference processes of AI models, with the foundational support layer abstracting hardware complexity and providing standard interfaces for upper-layer tools and applications [17][19] Group 3: Current State of Domestic AI Chip Software Ecosystem - The report identifies various domestic AI chip categories and representative manufacturers, highlighting their unique software ecosystem resources and the need for improvement in completeness and community engagement [3][10] - The current ecosystem has progressed from "basic usability" to "specific scenario usability," with two main paths emerging: "full-stack ecosystem" and "compatible ecosystem" [1][10] Group 4: Future Development Directions - The report emphasizes the need for standardization, open-source collaboration, and industry-academia synergy to enhance the domestic AI chip software ecosystem from "usable" to "excellent" [1][10] - The ultimate goal is to establish a robust and controllable technological system that supports the commercialization of domestic AI chips [1][10]

Core Insights - The article highlights a significant shift in the procurement strategy of AI computing power by top universities, moving from traditional hardware specifications to practical performance and compatibility with mainstream AI models like DeepSeek and Qwen [1][10]. Procurement Strategy - The procurement document emphasizes the importance of practical performance and compatibility over mere hardware specifications, marking a departure from the previous focus on "parameter stacking" [3][10]. - The core requirements include a focus on domestic production, ensuring real-world performance, and compatibility with existing CUDA ecosystems [3][10]. Technical Specifications - The procurement specifies the need for 13 GPU servers, with detailed requirements for CPU, memory, storage, and network capabilities, emphasizing the use of domestic components [4][6][8]. - Key performance indicators include the ability to run specific AI models under defined conditions, such as maintaining low latency and high throughput during inference tasks [9][11]. Trends in AI Computing Power Procurement - The article identifies three major trends in GPU server procurement by universities: 1. Transitioning from "indicator comparison" to "model testing" [10]. 2. Moving from a "NVIDIA-dominated ecosystem" to a "domestic compatible ecosystem" [10]. 3. Shifting focus from "hardware procurement" to "computing power system construction" [11]. Market Implications - The procurement strategy indicates a growing demand for domestic AI computing solutions that can effectively replace imported technologies, thereby fostering the development of a robust domestic AI ecosystem [12]. - Companies like Haiguang, Biran, and Muxi are highlighted as potential suppliers capable of meeting these stringent requirements, showcasing advancements in their GPU and CPU technologies [11][12].

Investment Rating - The report maintains a "Buy" rating for the company [1][11]. Core Insights - The company reported a revenue growth of 54.65% year-on-year for the first three quarters of 2025, reaching 9.49 billion yuan, with a net profit of 1.96 billion yuan, up 28.56% year-on-year [5][10]. - In Q3 2025, the revenue accelerated to 4.03 billion yuan, marking a 69.60% increase year-on-year, while the net profit was 760 million yuan, up 13.04% year-on-year [5][10]. - The company is expanding its market presence through deepened collaborations with OEMs and ecosystem partners, particularly in key industries [5]. Revenue and Profitability - The gross margin for the first three quarters of 2025 was 60.10%, a decrease of 5.54 percentage points year-on-year [6]. - The company has significantly increased its R&D and sales expenses, with R&D expenses rising by 42.55% year-on-year to 2.59 billion yuan [6]. - The net cash flow from operating activities reached 2.26 billion yuan, a remarkable increase of 465.64% year-on-year [6]. Market Potential - The company’s CPU and DCU segments are expected to experience substantial growth, particularly with increasing demand from government, internet, and financial sectors [7]. - The company is positioned as a leader in the domestic high-end chip market, benefiting from the accelerated development of AI computing power [10]. Financial Forecast - Revenue projections for 2025-2027 are 16.84 billion yuan, 27.19 billion yuan, and 42.74 billion yuan, respectively, with net profits of 3.90 billion yuan, 6.38 billion yuan, and 11.43 billion yuan [10][11]. - The estimated EPS for 2025-2027 is 1.68 yuan, 2.75 yuan, and 4.92 yuan, with corresponding P/E ratios of 138.74 and 47.39 for 2026 and 2027 [10][11].

Investment Rating - The report maintains a "Buy" rating for the company, indicating an expected relative performance increase of over 15% against the benchmark index within the next 6 to 12 months [7] Core Views - The company has demonstrated strong revenue growth, with a projected increase in operating income from 6,012 million yuan in 2023 to 28,700 million yuan by 2027, reflecting a compound annual growth rate (CAGR) of approximately 39% [3][6] - The net profit attributable to the parent company is expected to rise from 1,263 million yuan in 2023 to 6,359 million yuan in 2027, with a consistent growth rate of around 40% [3][6] - The company is positioned in a high-demand sector, with significant orders reflected in a contract liability of 28 billion yuan as of the end of Q3 2025, indicating robust market demand [5] Summary by Sections Financial Performance - For the first three quarters of 2025, the company achieved revenue of 9,490 million yuan, a year-on-year increase of 54.65%, and a net profit of 1,961 million yuan, up 28.56% [4] - In Q3 2025 alone, revenue reached 4,026 million yuan, marking a 69.60% increase year-on-year and a 31.38% increase quarter-on-quarter [4] Growth Projections - Revenue projections for 2024, 2025, 2026, and 2027 are 9,162 million yuan, 13,760 million yuan, 20,600 million yuan, and 28,700 million yuan respectively, with growth rates of 52%, 50%, and 50% for the following years [3][6] - The expected earnings per share (EPS) are projected to grow from 0.54 yuan in 2023 to 2.74 yuan in 2027 [3][6] Market Position and Strategy - The company is actively expanding its market presence and investing in research and development, with R&D expenses increasing by 42.55% year-on-year in the first three quarters of 2025 [5] - The opening of the HSL protocol is expected to enhance collaboration across the industry, improving efficiency and resource utilization, which will further drive innovation and market growth [5]

Investment Rating - The investment rating for the company is "Buy" (maintained) [2] Core Views - The company reported a significant increase in revenue and net profit for the third quarter of 2025, with a year-on-year revenue growth of 69.60% and a net profit growth of 13.04% [7] - The company is focusing on ecological construction and expanding its customer base in the internet sector, which is reflected in a substantial increase in sales and R&D expenses [7] - A stock incentive plan has been announced, demonstrating the company's long-term confidence in its growth potential [7] - The company is expected to maintain its leading position in the computing power sector, with projected revenues of 144.70 billion, 227.60 billion, and 338.50 billion for 2025, 2026, and 2027 respectively [7] Financial Performance Summary - For the first three quarters of 2025, the company achieved a revenue of 94.90 billion, a year-on-year increase of 54.65%, and a net profit of 19.61 billion, up 28.56% year-on-year [7] - The company’s revenue forecast for 2025 is 144.70 billion, with a net profit of 27.69 billion, reflecting a net profit growth rate of 43.41% [6][7] - The company’s R&D investment reached 12.24 billion in Q3 2025, a year-on-year increase of 53.83% [7] Stock Incentive Plan - The company has launched a stock incentive plan for 2025-2027, granting up to 2,068,430 restricted shares at a price not lower than 90.25 yuan per share, which accounts for approximately 0.89% of the total share capital [7] - The revenue growth targets for the stock incentive plan are set at 50%, 90%, and 140% for the years 2025, 2026, and 2027 respectively [7] Market Performance - The company has outperformed the market with a 3Q2025 revenue growth of 69.60%, compared to the broader market indices [4][7]

Core Viewpoint - Haiguang Information (688041) has reported strong financial results for Q3, marking the beginning of the disclosure season for the Shanghai Stock Exchange's Sci-Tech Innovation Board, with significant revenue and profit growth driven by strategic partnerships and product expansion [2] Financial Performance - In Q3, the company achieved revenue of 4.026 billion yuan, a year-on-year increase of 69.60%, and a net profit attributable to shareholders of 760 million yuan, up 13.04% [2] - For the first three quarters, total revenue reached 9.49 billion yuan, reflecting a 54.65% year-on-year growth, while net profit attributable to shareholders was 1.961 billion yuan, an increase of 28.56% [2] Product and Market Strategy - Haiguang Information has deepened collaborations with OEMs and ecosystem partners in key industries, accelerating client adoption and expanding the market presence of high-end processor products [2] - The company's main products include high-end general-purpose processors (CPUs) and co-processors (DCUs), with a focus on building a comprehensive software stack to support self-developed operators and third-party components [2] Industry Trends - The rise of generative AI technology has led to a revolution in AI acceleration processors, positioning Haiguang Information to benefit from the domestic AI cluster era [2] - The company's CPUs are compatible with the x86 instruction set, allowing effective integration with millions of existing software applications, enhancing its ecosystem advantage [3] Research and Development - R&D investment has significantly increased, with 2.935 billion yuan spent in the first three quarters, a 35.38% year-on-year growth, and 1.224 billion yuan in Q3 alone, up 53.83% [4] - The company is focused on enhancing product competitiveness through accelerated R&D in next-generation general-purpose processor designs and key technology developments [4] Strategic Initiatives - Haiguang Information is actively pursuing a merger with Zhongke Shuguang to leverage complementary strengths in chip design and data center infrastructure, aiming for vertical integration and technological collaboration [3]

Core Viewpoint - Haiguang Information has achieved record high revenue and net profit in Q3, driven by deepened cooperation with key industry partners and significant market expansion of high-end processor products [2][3]. Financial Performance - Q3 revenue reached 4.026 billion yuan, a year-on-year increase of 69.6% and a quarter-on-quarter increase of 31.38% [2]. - Net profit attributable to shareholders for Q3 was 760 million yuan, up 13.04% year-on-year and 9.26% quarter-on-quarter; for the first three quarters, net profit totaled 1.961 billion yuan, a 28.56% increase year-on-year [3]. - Operating cash flow increased by 465.64% compared to the beginning of the year, attributed to rapid business growth and increased sales collections [3]. Research and Development - R&D investment for the first three quarters grew by 35.38% [4]. Incentive Plan - In September, Haiguang Information announced a draft for a restricted stock incentive plan aimed at boosting confidence in domestic computing power shipments, focusing on revenue from CPU and DCU products [5]. - The performance targets set in the plan include a revenue base of 9.162 billion yuan for 2024, with growth rates of no less than 55%, 125%, and 200% for the following years [6]. Ecosystem Development - Haiguang Information opened the CPU interconnect bus protocol (HSL) in September, which is expected to enhance collaboration across the industry, improve computational efficiency, and promote the establishment of unified standards [7]. - Analysts suggest that the HSL protocol positions Haiguang CPU to play a crucial role in the evolving landscape of heterogeneous computing, similar to NVIDIA's architecture [7]. Inventory and Market Strategy - As of the end of Q3, contract liabilities remained high at 2.8 billion yuan, with increased prepayments and inventory levels indicating a proactive approach to supply chain management [8]. - The company aims to deepen collaborations with existing clients while exploring new market opportunities to expand its market share [8]. Stock Performance - As of the latest close, Haiguang Information's stock rose by 5.91%, with a market capitalization exceeding 560 billion yuan [9].

Core Insights - AMD has signed a 6GW computing power chip cooperation agreement with OpenAI, which is expected to accelerate the ROCm ecosystem development [1] - The Haiguang DTK software stack is fully compatible with the ROCm ecosystem, potentially reducing the adaptation costs for large models in China [2] - Industrial and commercial banks, along with China Unicom, are procuring Haiguang chips worth several billion yuan, indicating a rapid penetration of Haiguang chips in the domestic market [3] Group 1 - AMD and OpenAI's agreement includes the deployment of a total of 6GW of AMD GPU computing power, starting with 1GW based on the MI450 series GPU in the second half of 2026 [1] - The collaboration is seen as a response to the significant gap in AI computing power supply and aims to enhance the synergy between AMD and OpenAI in both hardware and software [1] - ROCm 7 is expected to be available by Q3 2025, offering over 3.5 times the inference capability and three times the training capability compared to ROCm 6 [1] Group 2 - The Haiguang DCU features a complete software stack, including DTK, development toolchain, and model repository, which supports mainstream deep learning frameworks like TensorFlow and Pytorch [2] - This compatibility is anticipated to effectively lower the costs associated with adapting large models domestically [2] Group 3 - Industrial and Commercial Bank of China has announced a procurement project for Haiguang chip servers, with a potential total procurement amount reaching 3 billion yuan [3] - China Unicom's procurement project for Haiguang servers has a budget of 2.782 billion yuan, indicating strong demand for domestic chip solutions [3] - The financial and telecommunications sectors are expected to lead the way in achieving comprehensive domestic production, thereby accelerating the penetration of Haiguang chips [3]

Core Viewpoint - The release of DeepSeek-V3.2-Exp model by DeepSeek Company marks a significant advancement in the domestic AI chip ecosystem, showcasing a collaborative effort among various domestic chip manufacturers [1][4][7]. Group 1: Model Release and Features - DeepSeek-V3.2-Exp introduces DeepSeek Sparse Attention, which significantly reduces computational resource consumption and enhances inference efficiency [1][7]. - The new model has led to a price reduction of API services by 50% to 75% across DeepSeek's platforms [1]. - The model's release prompted immediate recognition and adaptation from several domestic chip manufacturers, including Cambrian, Huawei, and Haiguang [2][4]. Group 2: Industry Response and Ecosystem Development - Cambrian was the first to announce compatibility with DeepSeek-V3.2-Exp, followed by Huawei and Haiguang, indicating a rapid response from the industry [2][4]. - The consensus within the domestic AI industry regarding DeepSeek's models has enabled the company to take the lead in defining standards for domestic chips [4][7]. - The rapid adaptation of DeepSeek's models by various manufacturers suggests a growing synergy within the domestic AI hardware and software ecosystem [9]. Group 3: Future Implications - Experts believe that the swift development of domestic chips by 2025 can be attributed to the emergence of DeepSeek as a key player in the industry [4][5]. - The collaborative efforts among domestic companies to adapt to DeepSeek's standards may accelerate the growth of the AI chip ecosystem in China [4][9]. - The advancements made by DeepSeek in a short time frame highlight the potential for rapid evolution in the domestic AI landscape, contrasting with the decades-long establishment of ecosystems by companies like NVIDIA [9].

Core Insights - The release and open-sourcing of the DeepSeek-V3.2-Exp model on September 29 marks a significant advancement in AI technology, featuring a sparse Attention architecture that reduces computational resource consumption and enhances inference efficiency [1] - The model's API pricing has been reduced by over 50%, making it more accessible for developers and users [1] - Major companies such as Huawei, Cambricon, and Haiguang have announced successful adaptations of the DeepSeek-V3.2-Exp model, showcasing its compatibility and performance across different platforms [1] Group 1 - The DeepSeek-V3.2-Exp model introduces a sparse Attention mechanism that significantly lowers training and inference costs in long-sequence scenarios [1] - Huawei's Ascend has quickly adapted the model for deployment, providing open-source inference code and operator implementations for developers [1] - Cambricon has also achieved adaptation of the DeepSeek-V3.2-Exp model, leveraging its computing efficiency to enhance performance [1] Group 2 - Following the announcement, the market saw a strong performance in the Sci-Tech 50 Index, with notable gains in AI chip stocks and Huawei Ascend concepts [2] - The dual advancement of hardware and software signifies a shift in the domestic AI ecosystem from "usable" to "user-friendly," creating a closed loop from foundational computing power to upper-layer applications [2] - Analysts suggest that the integration of large models and generative AI into consumer endpoints will enhance product value and user engagement, positioning companies that develop "AI + hardware" applications competitively in the next computing platform race [2]