（文章来源：证券日报） 证券日报网讯 2月3日，紫光股份在互动平台回答投资者提问时表示，公司2025年度经营情况请详见公 司拟定于2026年4月29日披露的《2025年年度报告》。 ...

Core Viewpoint - Unisplendour Co., Ltd. (紫光股份) is actively involved in the sales of Western Digital hard drive products and SanDisk SSD products in China through its subsidiary, Unisplendour Western Data Co., Ltd. [1] Group 1 - The main business of Unisplendour Western Data Co., Ltd. includes the sale of various hard drive products in China [1] - The product offerings consist of high-capacity enterprise-grade mechanical hard drives, consumer-grade mechanical hard drives, enterprise-grade solid-state drives (SSDs), and consumer-grade SSDs [1]

紫光股份（000938.SZ）2月3日在投资者互动平台表示，公司2025年度经营情况请详见公司拟定于2026 年4月29日披露的《2025年年度报告》。 （记者 王晓波） 每经AI快讯，有投资者在投资者互动平台提问：公司是不是有业绩2025年预告？ ...

Group 1 - The core storage supply chain is facing a structural shortage that is expected to persist until 2027, driven by the increasing demands of generative AI and the need for high bandwidth and large capacity GPU memory [2] - The reliance on hardware stacking to address these challenges is not sustainable, leading to increased costs per token and affecting the healthy development of the AI industry [2] - Optimizing the efficiency of key components like GPUs through software and hardware collaboration is essential to mitigate supply chain shortages and reduce overall ownership costs [2] Group 2 - The collaboration between Unisplendour and Pliops aims to address the exponential growth in computational and memory demands of large model inference, which is constrained by the costs and energy efficiency of stacking GPU hardware [3] - The Pliops custom ASIC chip offloads KV Cache from GPU memory to dedicated storage nodes, creating a new memory layer designed for AI, thus alleviating GPU memory pressure [3][4] - This solution supports both single-machine deployment and external storage nodes to enhance the inference performance of multiple AI servers [4] Group 3 - Benchmark tests conducted by Unisplendour on their high-performance AI servers show significant improvements in inference performance when using the KV Cache offloading acceleration scheme [7] - The number of concurrent users supported increased by 200%, while the first token generation delay (TTFT) decreased by 70% and the average token generation delay (TPOT) reduced by 30% [7] Group 4 - The solution is adaptable for various enterprise GenAI applications, including interactive applications like chatbots and intelligent customer service, where quick loading of historical KV Cache enhances user experience [8] - It also effectively handles long-context tasks that require processing thousands of tokens, providing PB-level KV Cache expansion capabilities to avoid performance degradation due to GPU memory limitations [8] - The efficient management of KV Cache allows for higher throughput in online inference services, enabling the system to handle more concurrent requests with the same GPU resources [8] Group 5 - The continuous innovation in inference acceleration is crucial for the future of AI infrastructure, with Unisplendour committed to developing tailored acceleration solutions for various scenarios to help enterprises and developers tackle the complexities of large model applications [9]

Core Viewpoint - The report by Zhiyan Consulting highlights the trends and forecasts for the optoelectronic device industry in China from 2026 to 2032, indicating a slight decline in production in December 2025 but an overall growth in annual production for the year [1] Group 1: Industry Overview - In December 2025, China's optoelectronic device production reached 173.3 billion units, reflecting a year-on-year decrease of 0.6% [1] - For the entire year of 2025, the cumulative production of optoelectronic devices in China was 1,923.39 billion units, showing a cumulative growth of 8.8% [1] Group 2: Market Research and Insights - Zhiyan Consulting is recognized as a leading industry consulting firm in China, specializing in in-depth industry research reports, business plans, feasibility studies, and customized services [1] - The firm emphasizes its commitment to providing comprehensive industry solutions to empower investment decisions through professional insights and quality services [1]

Summary of Conference Call Records Industry and Company Involved - The discussion primarily revolves around the **SaaS (Software as a Service)** industry, particularly focusing on the impact of AI advancements on software companies in both the US and China. - The emergence of **MoltBook**, an AI Agent social network developed from **Clawdbot**, is highlighted as a significant development in the industry. Core Points and Arguments 1. **Launch of MoltBook**: - On January 28, MoltBook was launched, attracting over **150,000 AI Agents** and creating more than **12,000 submolt communities** within a few days, indicating rapid adoption and engagement in the AI Agent ecosystem [1][1][1]. 2. **Decline in US SaaS Sector**: - The US SaaS sector experienced a significant downturn, with companies like **Unity** leading the decline. Previous adjustments in stock prices were noted for **SAP**, **ServiceNow**, and **Salesforce**. Concerns were raised about "large models consuming software and SaaS" [1][1][1]. 3. **Impact of AI on Software Companies**: - The US software industry has successfully transitioned to SaaS over the past decade, focusing on process management and collaboration. However, the simplicity of these systems makes them vulnerable to disruption by AI Agents [1][1][1]. 4. **Chinese SaaS Landscape**: - In China, major internet companies dominate the SaaS market (e.g., **Feishu**, **DingTalk**), while listed software firms focus on vertical industry solutions and primarily offer private deployments. This suggests that the impact of AI models and Agent startups on Chinese software companies will be delayed and less severe compared to the US [2][2][2]. 5. **Complexity of Core Systems**: - The complexity and specialization of core systems across industries are expected to remain resilient against disruption by Agents in the short term. Companies are encouraged to accelerate their transformation by integrating Agent-related technologies to meet future AI demands [3][3][3]. 6. **Opportunities in Computing and Storage**: - The demand for computing and storage solutions is projected to grow positively, with opportunities in AI databases and storage being particularly promising. Companies in CPU, cloud computing, and AI database sectors are recommended for investment [4][4][4]. 7. **Investment Recommendations**: - Suggested companies for investment include: - **StarRing Technology** in the database sector - **Haiguang Information** and **Hesheng New Materials** in the CPU sector - **Runjian Co.**, **Beijing Online**, **UCloud**, and others in the computing service sector - **Sangfor Technologies**, **Pinggao Group**, and others in the private cloud sector - **Inspur Information** and **Unisplendour** in the server sector [5][5][5]. Other Important but Overlooked Content - The security risks associated with the rise of Agents are acknowledged, emphasizing the need for comprehensive solutions to address these challenges [4][4][4].

Core Insights - The rise of Artificial Intelligence Data Centers (AIDC) has transformed electricity from a marginal cost to a core resource, driving companies to seek dual financing platforms in the capital market [1][4] - Companies like Jinpan Technology and Deye Co., Ltd. are moving towards internationalization through IPOs in Hong Kong, reflecting a shift in capital strategy from "selling computing power" to "controlling energy" [2][3] - The demand for high-efficiency power distribution and solid-state transformers (SST) is increasing, with Jinpan Technology reporting a 603.68% year-on-year growth in data center orders for 2024 [3][5] Industry Dynamics - The AIDC industry is undergoing a systemic reconstruction, moving from a focus on "computing power chips" to "full-link infrastructure," with significant advancements in high-voltage direct current (HVDC) technology [5][9] - The Chinese power consumption is projected to exceed 10 trillion kilowatt-hours by 2025, indicating a doubling in ten years, which is more than twice that of the United States [5][6] - The liquid cooling revolution is becoming a critical component in data centers, with significant cost increases for cooling solutions, highlighting the rising demand for efficient thermal management [6][8] Capital Market Trends - The collective move of companies to list in Hong Kong is not merely for financing but aims to establish a strategic loop for global supply chain anchoring and liquidity premium [4][8] - The integration of AI infrastructure with industrial applications is emphasized in recent government policies, indicating a shift towards a more collaborative and efficient AI ecosystem [7][8] - The capital expenditure of major North American AI companies is expected to rise significantly, with projections showing an increase from $24.1 billion in Q2 2023 to $76 billion by Q3 2025 [8][9] Future Outlook - The AIDC industry is anticipated to experience significant changes by 2026, with expectations of breakthroughs in computing power, models, and applications [9] - Companies that can effectively address "energy anxiety" will gain a competitive edge in redefining industry dynamics [9]

Core Insights - The rise of Artificial Intelligence Data Centers (AIDC) is transforming the energy landscape, making electricity a critical resource for computing power limits [1][2] - AIDC companies are rapidly crossing capital boundaries, initiating a new era of dual financing platforms, referred to as "A+H" [3] - The shift from "selling computing power" to "controlling energy" reflects a deep transformation in industry logic, with companies racing to secure funding before global energy shortages become a consensus [4] Company Developments - Jinpan Technology, a leading supplier of AIDC power equipment, reported a staggering 603.68% year-on-year increase in data center orders for 2024 [7] - The company is expanding its R&D into high-voltage direct current (HVDC) and solid-state transformers (SST), with expected revenue of approximately 5.194 billion yuan in the first three quarters of 2025, a year-on-year increase of 8.25% [7] - DeYe Co., a leader in household energy storage inverters, is set to submit its IPO application to the Hong Kong Stock Exchange in January 2026, focusing on SST and smart cloud platforms [7] - Unisplendour is leveraging its high market share in networking to address communication latency issues in AI clusters [7] Market Trends - The collective move to the Hong Kong market is not merely a financing action but a strategic effort to anchor global supply chains and enhance liquidity premiums [8] - The AIDC industry is undergoing a systemic reconstruction, transitioning from a focus on computing chips to comprehensive infrastructure [11] - The demand for efficient power supply systems is increasing, with traditional AC power being replaced by more efficient 800V HVDC systems [12] Policy and Future Outlook - Recent government policies indicate a shift from "computing power infrastructure" to "industrial deep integration," emphasizing the need for enhanced AI computing supply [18] - The focus on "artificial intelligence+" actions highlights the importance of iterative technological breakthroughs driven by application demands [19] - The anticipated growth in capital expenditures from North American AI giants and Chinese internet companies signals a robust demand for AIDC infrastructure [20]

[Table_Title] 计算机行业 AI 2026 算力系列（二） 从云业务到千问 APP，阿里算力需求保持 旺盛态势 [Table_Summary] 核心观点： [Table_Page] 跟踪分析|计算机 证券研究报告 | [Table_Grade] 行业评级 | 买入 | | --- | --- | | 前次评级 | 买入 | | 报告日期 | 2026-01-30 | [Table_PicQuote] 相对市场表现 -14% -4% 5% 15% 24% 34% 02/25 04/25 06/25 09/25 11/25 01/26 计算机 沪深300 | [分析师： Table_Author]刘雪峰 | | --- | | SAC 执证号：S0260514030002 | | SFC CE No. BNX004 | | 021-38003675 | | gfliuxuefeng@gf.com.cn | | 分析师： 周源 | | SAC 执证号：S0260523040001 | | 0755-23948351 | | shzhouyuan@gf.com.cn | | 请注意，周源并非香港证券及期货 ...

Core Viewpoint - The semiconductor industry chain experienced a rapid increase due to positive news regarding domestic chips, with significant gains in stocks such as 澜起科技, which rose over 11% [1]. Group 1: Market Performance - The AI ETF (515070) saw a recovery in the afternoon after an early decline of over 2%, ultimately rising more than 1.2% with a trading volume exceeding 2 billion yuan [1]. - Stocks in the semiconductor sector, including 新易盛 and 北京君正, also experienced notable increases, reflecting a broader market rally [1]. Group 2: AI Developments - Baidu's Wenxin APP launched the industry's first "multi-agent" group chat beta, allowing multiple specialized agents to collaborate, enhancing AI's role as a "smart collaborator" in decision-making processes [1]. - This advancement signifies a shift in AI interaction from "human-machine dialogue" to "human-machine group chat," indicating a more integrated approach to AI in collaborative environments [1]. Group 3: Market Outlook - Dongwu Securities noted that favorable factors driving market strength remain unchanged, suggesting a stable transition into the second phase of the spring market rally [1]. - The China Securities Regulatory Commission emphasized maintaining stability and supporting a gradual market recovery, which is expected to provide a solid foundation for a sustained bull market [1]. - As the annual performance forecasts begin to be disclosed in late January, market focus is anticipated to shift towards earnings reports [1]. Group 4: AI ETF Composition - The AI ETF (515070) tracks the CS AI theme index (930713), selecting stocks that provide technology, foundational resources, and applications within the AI industry, often referred to as the "brain" creators of robotics and the "infrastructure" of the Internet of Things [2]. - The top ten weighted stocks in the ETF include leading domestic technology firms such as 中际旭创, 新易盛, and 科大讯飞, among others [2].