证券研究报告 计算机 报告日期：2026 年 03 月 23 日 Minimax 发布 M2.7，阿里云部分产品提价 ——计算机行业周报 华龙证券研究所 投资评级：推荐（维持） 最近一年走势 执业证书编号：S0230526030001 邮箱：zhulx@hlzq.com 《政府工作报告再提"人工智能+"， 2026.03.09 AI 基建投资机会—计算机行业周报》 2026.03.02 《北美云服务局部涨价，AI 应用按下加 速键—计算机行业周报》2026.02.02 请认真阅读文后免责条款 摘要： 执业证书编号：S0230523080004 邮箱：sunbw@hlzq.com 分析师：孙伯文 MiniMax 发布 M2.7，首次实现模型自我进化能力。MiniMax 于 2026 年 3 月 18 日发布了新一代 Agent 旗舰大模型 M2.7，核心的 突破在于首次实现了"模型自我进化" 。M2.7 能够通过构建复 杂 Agent Harness，深度参与自身的训练、优化与迭代流程，例如 自动执行超过 100 轮"分析-改进-验证"的闭环，最终通过优化 在内部评测集上实现约 30%的效果提升，并能够在 RL ...

Investment Rating - The report maintains a "Recommended" rating for the computer industry [1] Core Insights - The demand for AI inference is continuously growing, driven by a significant increase in tokens usage, with a reported increase from 1.62 trillion tokens in March 2025 to 18 trillion tokens in March 2026, representing a growth of approximately 1011% [6][11] - Cloud service providers are raising prices for AI computing products due to rising hardware costs and increased demand for AI services, with price hikes ranging from 5% to 34% for various services [8][33] - The report highlights that the expansion of AI capabilities is expected to lead to a substantial increase in the number of active agents and token consumption, with annual token consumption projected to grow from 0.0005 PetaTokens in 2025 to 152,667 PetaTokens by 2030, reflecting a compound annual growth rate of 3418% [9][38] Summary by Sections Recent Trends - The computer industry has shown a relative performance decline of -10.1% over the past month, while the Shanghai Composite Index has remained stable [5] Price Adjustments - Major cloud providers like Alibaba Cloud and Baidu Cloud are increasing prices for AI computing services due to rising hardware procurement costs, with specific increases of up to 34% for Alibaba's AI computing services and 30% for Baidu's [8][27] Token Consumption Growth - The report indicates a significant rise in token consumption, with OpenClaw being a major contributor, achieving a monthly token call volume of 13.4 trillion as of March 2026 [17] - The share of domestic models in token consumption is increasing, with domestic models accounting for approximately 53.4% of the top models' total token calls as of March 2026 [14] Future Outlook - The report anticipates that the demand for AI computing and tokens will continue to rise, benefiting cloud service providers and related upstream and downstream companies [10][44] - The ongoing increase in hardware costs and the demand for AI services suggest that price adjustments in the cloud computing sector may persist [9][33]

股票研究 /[Table_Date] 2026.03.22 [Table_Industry] 计算机 计算机周观点第 37 期：大模型进入可执 行 Agent 时代，入口与算力侧同步演进 | [姓名table_Authors] | 电话 | 邮箱 | 登记编号 | | --- | --- | --- | --- | | 杨林(分析师) | 021-23183969 | yanglin2@gtht.com | S0880525040027 | | 吕浦源(分析师) | 021-23183822 | lvpuyuan@gtht.com | S0880525050002 | | 钟明翰(研究助理) | 021-38031383 | zhongminghan@gtht.com | S0880124070047 | 本报告导读： 本周，小米、MiniMax、Anthropic 及腾讯 QClaw 继续强化模型 Agent 能力与入口 布局，英伟达重启 H200 对华生产、新一代机架"光铜并举"，产业演进持续提速。 投资要点： [Table_Invest] 评级： 增持 计算机《计算机周观点第 36 期：OpenClaw ...

海光信息技术股份有限公司 2026年第一次临时股东会会议资料 海光信息技术股份有限公司 2026 年第一次临时股东会 会议资料 二〇二六年三月 | | | 海光信息技术股份有限公司 2026年第一次临时股东会会议资料 海光信息技术股份有限公司 2026 年第一次临时股东会 会议须知 根据中国证券监督管理委员会《上市公司股东会规则》和《海光信息技术股 份有限公司章程》的有关规定，为便于各位股东及其授权代理人出席会议并确保 会议顺利进行，以下事项敬请注意： 一、本次会议期间，全体参会人员应以维护股东的合法权益，确保会议的正 常秩序和议事效率为原则，自觉履行法定义务。 二、请拟参加本次股东会现场会议的股东按时办理会议登记手续，会议登记 方法详见本须知第四项。请确认与会资格的股东准时出席股东会现场会议。 三、为保证本次会议的严肃性和正常秩序，除出席会议的股东及股东代理人、 董事、公司高级管理人员、公司聘任律师及中介机构外，公司有权依法拒绝其他 人员入场。 四、会议登记方法： （一）参加股东会现场会议的登记时间、地点 时间：2026 年 3 月 25 日上午 9:00-11:00，下午 14:00-16:00 地点：北 ...

Group 1 - The report maintains a positive outlook on the AI industry chain for 2026, highlighting the significant growth in the computer sector in 2025, driven by AI trends and applications [2][25][27] - The computer sector's revenue accelerated, with a notable improvement in net profit, indicating a potential for high valuations to be gradually digested [16][17][21] - The demand for AI computing power remains robust, with domestic chip manufacturers making significant breakthroughs in performance, ecosystem, and production capacity [3][27][35] Group 2 - C-end AI applications have rapidly developed, with major internet companies dominating the market, while B-end applications are expected to see explosive growth in 2026 [4][49][55] - The C-end market has seen a substantial increase in monthly active users, with AI applications achieving significant user engagement [52][55] - The report emphasizes the importance of high-barrier vertical applications in the B-end market, such as AI in industrial manufacturing and healthcare, which are anticipated to contribute significantly to revenue growth [4][25][49] Group 3 - The report recommends several stocks, including Haiguang Information, Kingsoft Office, and Zhongkong Technology, as key investment opportunities in the AI computing and application sectors [2][3][5] - The focus on domestic AI chip manufacturers is underscored, with companies like Huawei, Cambricon, and Haiguang Information being highlighted for their competitive advancements [3][35][45] - The report notes that the domestic AI chip market is expected to capture 70% of the high-end AI chip market share by 2026, significantly increasing the competitive landscape against international players [31][32][35]

Industry Investment Rating - The industry investment rating is "Outperform the Market" and is maintained [1] Core Insights - The report highlights the significant paradigm shift brought by OpenClaw, transitioning from a "dialogue tool" to a "digital employee," enabling autonomous execution of tasks through a four-layer architecture [4] - OpenClaw has sparked a nationwide technological craze, with its GitHub stars surpassing 270,000, leading to widespread adaptation by domestic manufacturers [4] - The demand for AI infrastructure is expected to grow exponentially due to OpenClaw's token consumption model, which has shifted from "human-machine dialogue" to "machine self-circulation" [4] Summary by Relevant Sections Industry Overview - The closing index for the industry is 5311.29, with a 52-week high of 6151.34 and a low of 4080.58 [1] Investment Recommendations - The report suggests focusing on: 1. Large models benefiting from API call surges: MiniMax, Zhipu, etc. 2. Instant Messaging (IM) entry points: Alibaba, Tencent, etc. 3. Cloud services as direct beneficiaries of user growth: UCloud, QingCloud, Wangsu Technology, etc. 4. Computing power leasing/IDC due to rising demand: Dazhi Technology, Dongyangguang, etc. 5. Computing hardware for foundational support: Cambricon, Haiguang Information, etc. [5]

Investment Rating - The industry investment rating is "Positive" (maintained) [1] Core Insights - OpenClaw, an open-source AI agent framework, has gained significant popularity since its release, becoming the most popular open-source project on GitHub as of February 2026. It offers capabilities such as local-first operation, autonomous tool invocation, cross-application execution, multi-IM access, and continuous online operation, making it highly appealing to users [4] - The demand for inference computing power is expected to grow exponentially due to the characteristics of Agent AI, which include autonomous task execution and multi-agent concurrency. The daily token consumption in China surged from 100 billion in early 2024 to over 30 trillion by mid-2025, reflecting a growth of over 300 times in just one and a half years [5] - OpenClaw is set to transform AI from a conversational role to an execution role, enabling more intelligent personal AI assistants and digital employees. Major companies like Xiaomi and Huawei are adapting OpenClaw for various devices, indicating a shift in the capabilities of edge AI [5] Summary by Sections OpenClaw Development - OpenClaw was initially released on November 24, 2025, and saw a surge in adoption in February 2026, with major domestic companies launching products based on its code. Notable implementations include Tencent's WorkBuddy and Zhizhu's AutoClaw, which offer seamless integration with existing communication tools [4] Token Consumption and Demand - The introduction of Agent AI has led to a shift in token consumption patterns, with tasks now consuming tens of thousands to millions of tokens per execution. Predictions indicate that by 2030, the global active AI agents will reach 2.216 billion, with annual token consumption skyrocketing from 0.0005 Peta Tokens in 2025 to 15.2 million Peta Tokens [5] Investment Recommendations - The report suggests focusing on the AI inference computing power supply chain, including segments like chips, complete machines, liquid cooling, and power supplies. Recommended stocks include Haiguang Information, Lingyi Technology, and Dongshan Precision, among others [6]

Investment Rating - The industry investment rating is "Positive" (maintained) [1] Core Insights - OpenClaw, an open-source AI agent framework, has gained significant popularity since its release, becoming the most popular open-source project on GitHub as of February 2026. It offers capabilities such as local-first operation, autonomous tool invocation, cross-application execution, and continuous online operation, making it highly appealing to users [4] - The demand for inference computing power is expected to grow exponentially due to the characteristics of Agent AI, which include autonomous task execution and multi-agent concurrency. The daily token consumption in China surged from 100 billion in early 2024 to over 30 trillion by June 2025, reflecting a growth of over 300 times in just one and a half years [5] - OpenClaw is penetrating various devices, including PCs, smartphones, and wearables, transforming the role of AI from a conversational agent to an executor. This shift is expected to create new application scenarios and product categories in the end-side AI market [6] Summary by Sections OpenClaw Development - OpenClaw has seen rapid adoption among major domestic companies, with multiple products and solutions being released based on its code. Notable companies include Tencent, Huawei, Alibaba, and Xiaomi, which are integrating OpenClaw into their cloud services and applications [4] Token Consumption and Inference Demand - The workflow of "plan-execute-feedback-replan" in Agent AI has led to a shift from linear to exponential growth in token consumption. A single task can consume tens of thousands to millions of tokens, indicating a significant increase in demand for inference capabilities [5] End-Side AI Transformation - The modular architecture of OpenClaw allows for comprehensive session management and memory systems, enabling it to autonomously manage various software applications. This evolution is expected to redefine the capabilities of end-side AI and lead to the development of more intelligent personal AI assistants [6] Investment Recommendations - The report suggests focusing on the AI inference computing power supply chain, including segments like chips, complete machines, liquid cooling, and power supplies. Recommended stocks include Haiguang Information, Lingyi Technology, and Dongshan Precision, among others [7]

Summary of Key Findings from China Semiconductors Tour 1H26 Industry Overview - The semiconductor industry in China is experiencing strong demand, particularly in sectors other than mobile, with companies gaining market share through technological advancements and localization efforts [1][2]. Core Insights Semiconductor Equipment (Semicap) - Front-end capacity expansion remains robust, with DRAM orders expected for a Shanghai fab soon and additional fabs planned for Beijing and Hefei by 2027-28 [2]. - Advanced logic orders are anticipated to bring significant upside, although visibility for mature logic remains less clear [2]. - The localization rate for semiconductor equipment is projected to rise from approximately 30% to 45% this year, with domestic suppliers gaining traction [12][26]. AI Chips - Demand for AI chips continues to be strong, but capacity constraints are a significant bottleneck [3]. - CSPs are shifting towards local vendors due to restrictions on Nvidia chips, increasing demand for local chips in inferencing [3]. Power Discrete - Foundry and IDM capacity in China is tightening, leading to expected price increases for power discrete components [4]. - Demand in the automotive sector is projected to grow over 20%, while consumer demand is expected to be weaker [4]. Analog Semiconductors - New product releases are driving market share gains, with pricing stabilizing and potential recovery expected this year [5]. - The automotive segment is anticipated to grow faster than others, with increasing content per vehicle [5]. Financial Guidance and Market Performance - Companies such as NAURA, AMEC, Piotech, SMIC, Hua Hong, Cambricon, Hygon, and Horizon Robotics are rated as outperformers, while Silergy is marked as market-perform and Black Sesame as underperform [8]. - AMEC aims for RMB 100 billion in revenue over the next five years, targeting a 30% CAGR [23]. - The overall gross margin for semicap is expected to decline to around 39% in 2026, driven by competition and margin compression in mature-node and memory segments [18]. Investment Implications - The semiconductor equipment market in China is projected to reach RMB 500-600 billion by 2030, with a conservative estimate of capturing 20% market share [23]. - The localization of semiconductor equipment is seen as irreversible, with significant opportunities for domestic suppliers [12][13]. Additional Observations - The competitive landscape is characterized by regional government support and competition, which is driving growth in the semiconductor sector [26]. - Companies are focusing on R&D and product development to enhance competitiveness, particularly in advanced technology areas such as GAA and 3D NAND processes [30][41]. Conclusion - The semiconductor industry in China is poised for growth, driven by strong demand across various segments, localization efforts, and technological advancements. Companies are strategically positioning themselves to capitalize on these trends, with a focus on expanding market share and improving financial performance.

Investment Rating - The report maintains an "Overweight" rating for the computer sector [4] Core Insights - The OpenClaw ecosystem is rapidly expanding, with deployment barriers decreasing and competition intensifying. OpenAI's release of GPT-5.4 enhances model capabilities towards a unified approach. AWE 2026 serves as a significant launch window for new domestic chips, indicating a shift towards infrastructure and cutting-edge applications [2][5] Summary by Sections Investment Highlights - The OpenClaw ecosystem is experiencing rapid growth, with a 30-day token usage of approximately 10.2 trillion tokens and coverage of 344 models as of March 13. Domestic companies are intensifying their deployment and entry strategies, with notable developments from Zhiyu and other major players [5] - OpenAI's GPT-5.4 integrates reasoning, programming, native computer interaction, deep web search, and supports a context of millions of tokens, achieving success rates of 67.3%, 92.8%, and 75.0% in various tests. The model also reduces token usage by 47% while lowering factual error rates by 33% [5] - AWE 2026 is highlighted as a crucial platform for launching new domestic chips, with various companies showcasing advancements in GPU technology and smart home computing solutions [5] Recommended Stocks - The report recommends the following stocks with their respective earnings forecasts and ratings: - Rilian Technology: EPS forecast of 1.25 in 2024, rated "Overweight" [6] - Kingsoft Office: EPS forecast of 3.56 in 2024, rated "Overweight" [6] - Haiguang Information: EPS forecast of 0.83 in 2024, rated "Overweight" [6] - Inspur Information: EPS forecast of 1.56 in 2024, rated "Overweight" [6] - Hehe Information: EPS forecast of 4.93 in 2024, rated "Overweight" [6] - Hikvision: EPS forecast of 1.30 in 2024, rated "Overweight" [6] - Saiyi Information: EPS forecast of 0.34 in 2024, rated "Overweight" [6] - Xinguodu: EPS forecast of 0.42 in 2024, rated "Overweight" [6] - Xunce: EPS forecast of -0.28 in 2024, rated "Overweight" [6] - Jushuitan: EPS forecast of 6.82 in 2024, rated "Overweight" [6]