Group 1 - The core point of the article is that Softstone Power (301236.SZ) announced a share reduction plan by its controlling shareholder Liu Tianwen and his action group, Softstone Intelligent Venture Capital Partnership [1] - Liu Tianwen holds 226,354,751 shares, accounting for 23.75% of the total share capital, while Softstone Intelligent holds 46,271,037 shares, representing 4.86% of the total share capital [1] - The planned reduction involves selling up to 19,058,822 shares, which is no more than 2% of the total share capital, through a combination of centralized bidding and block trading within three months after the announcement [1]

Core Viewpoint - The controlling shareholder Liu Tianwen and his action-in-concert party, Xiong'an Ruoshi Zhihua Investment Partnership, plan to reduce their shareholding in Softcom Power Information Technology Group Co., Ltd. due to personal financial arrangements [1][2]. Summary by Sections Reduction Plan Details - As of the announcement date, Liu Tianwen holds 226,354,751 shares, accounting for 23.75% of the total share capital, while Soft Stone Zhihua holds 46,271,037 shares, accounting for 4.86% [2]. - They plan to reduce up to 19,058,822 shares, representing 2% of the total share capital, within three months from October 21, 2025, to January 20, 2026, through a combination of centralized bidding and block trading [2]. - The reduction will consist of a maximum of 9,529,411 shares via centralized bidding (1.00% of total share capital) and 9,529,411 shares via block trading (1.00% of total share capital) [2]. - The shares to be reduced are from those held prior to the initial public offering and from rights distribution, with the reduction price determined by market conditions at the time of sale [2]. Commitment Fulfillment - Liu Tianwen and Soft Stone Zhihua made commitments regarding shareholding and reduction intentions during the initial public offering [3]. - Liu Tianwen committed to a 36-month lock-up period post-listing, with annual transfer limits of 25% during his tenure and a six-month restriction post-departure [3]. - Soft Stone Zhihua also committed to a 36-month lock-up, with an extension of six months if the closing price falls below the issue price [3]. - Both parties have adhered to their commitments, and the current reduction aligns with previous promises [3].

Group 1 - The core point of the article is that Softstone Technology (301236.SZ) announced a plan for its controlling shareholder, Liu Tianwen, and his concerted party, to reduce their holdings in the company by up to 19.0588 million shares, which represents a maximum of 2% of the company's total share capital [1] Group 2 - The reduction in shareholding is planned to occur within three months after the announcement, starting 15 trading days from the date of the announcement [1] - The method of reduction will involve a combination of centralized bidding and block trading [1] - The reason for the reduction is attributed to the personal financial arrangements of the controlling shareholder and his concerted party [1]

人民财讯9月19日电，软通动力（301236）9月19日公告，公司近日收到控股股东、实际控制人刘天文及 其一致行动人软石智动的减持计划告知函。刘天文持有公司23.75%股份，软石智动持有4.86%股份。两 者计划通过集中竞价和大宗交易方式减持不超过1905.88万股，即不超过公司总股本的2%。 ...

证券代码：301236 证券简称：软通动力 公告编号：2025-081 软通动力信息技术（集团）股份有限公司 股东减持股份预披露公告 股东刘天文先生、雄安软石智动创业投资合伙企业（有限合伙）保证向本公 司提供的信息内容真实、准确、完整，没有虚假记载、误导性陈述或重大遗漏。 本公司及董事会全体成员保证公告内容与信息披露义务人提供的信息一致。 特别提示： 软通动力信息技术（集团）股份有限公司（以下简称"公司"或"软通动力"） 于近日收到控股股东、实际控制人刘天文先生及其一致行动人雄安软石智动创业 投资合伙企业（有限合伙）（以下简称"软石智动"）出具的《关于股份减持计 划的告知函》，刘天文先生持有公司股份数量 226,354,751 股，占公司总股本比 例 23.75%，软石智动持有公司股份数量 46,271,037 股，占公司总股本比例 4.86%。 刘天文先生、软石智动由于自身资金安排的原因，拟计划在本公告发布之日起 15 个交易日之后的 3 个月内，通过集中竞价和大宗交易相结合的方式减持不超 过 19,058,822 股公司股份，即减持不超过公司总股本的 2%。其中，通过集中竞 价方式减持不超过 9,529, ...

（文章来源：证券日报） 证券日报网讯 9月18日晚间，软通动力发布公告称，公司于近日接到股东刘天文先生函告，获悉其质押 了持有的公司部分股份，本次质押数量11，000，000股。 ...

Core Viewpoint - The stock of Softcom Power has experienced fluctuations, with a recent decline of 2.00% on September 19, 2023, and a total market capitalization of 53.231 billion yuan [1] Financial Performance - For the first half of 2025, Softcom Power achieved operating revenue of 15.781 billion yuan, representing a year-on-year growth of 25.99%, while the net profit attributable to shareholders was -143 million yuan, an increase of 7.60% year-on-year [2] - Since its A-share listing, Softcom Power has distributed a total of 584 million yuan in dividends over the past three years [3] Stock Market Activity - As of September 19, 2023, Softcom Power's stock price was 55.86 yuan per share, with a trading volume of 1.304 billion yuan and a turnover rate of 2.93% [1] - The stock has seen a year-to-date decline of 4.69%, a 5-day increase of 1.21%, a 20-day decline of 9.01%, and a 60-day increase of 3.25% [1] Shareholder Structure - As of June 30, 2025, the number of shareholders for Softcom Power was 134,900, a decrease of 4.89% from the previous period, with an average of 5,040 circulating shares per person, an increase of 5.14% [2] - The top ten circulating shareholders include notable entities such as E Fund's Growth Enterprise Board ETF and Hong Kong Central Clearing Limited, with varying changes in their holdings [3]

Core Insights - The mainstream narrative around artificial intelligence (AI) is undergoing a profound shift towards a new paradigm based on large models and agents as the core of interaction, accelerating penetration into various industries [2][4] - The AI industry is experiencing a valuation reconstruction, with significant interest from global investors in infrastructure-related stocks such as artificial intelligence, semiconductors, and computing chips [4][10] - The AI revolution is characterized as an "intelligent revolution," where AI evolves beyond being a mere tool to becoming intelligence itself, necessitating the emergence of "AI architects" across industries [5][7] Industry Trends - The demand for intelligent upgrades in sectors like finance, healthcare, manufacturing, and smart cities is surging due to the deep integration of large models [4][9] - The concept of "agent economy" is emerging, where economic activities will be coordinated and executed by agents, redefining labor markets and organizational structures [9][10] - The AI industry is expected to form a "dual-core" driving pattern, with the coexistence of closed-source and open-source large models, and the competition between the US and China as key players [10][11] Investment Opportunities - The AI sector is seen as a fertile ground for nurturing world-class companies, particularly in manufacturing and finance, with a focus on long-term investment strategies [8][9] - The construction of advanced computing infrastructure is critical for the development of artificial general intelligence (AGI), with a focus on creating more efficient and powerful computing centers [13][15] - Companies are encouraged to focus on vertical scenarios to create sustainable business models and address high-frequency pain points in specific industries [20][22] Technological Developments - The integration of AI into various sectors is leading to a transformation from human-centered services to agent-centered services, enhancing decision-making capabilities [19][20] - AI applications are expected to evolve from being productivity tools to becoming the core of productivity itself, emphasizing results over processes [19][22] - The AI hardware market is anticipated to thrive by combining agents with hardware and vertical scenarios, enhancing user experience through context-aware interactions [22][23] Educational Innovations - AI is poised to address traditional education challenges by providing personalized learning experiences and focusing on students' holistic development [25][29] - The integration of AI in education aims to overcome limitations such as teacher scarcity and uniform learning speeds, promoting tailored educational solutions [25][29]

Investment Rating - The report maintains an "Outperform" rating for the industry [6] Core Insights - Huawei's new products, including Ascend chips and supernodes, are set to lead a new era in computing power, with a clear roadmap for product iterations from 2025 to 2028 [1][14] - The introduction of self-developed HBM technology marks a significant advancement in memory bandwidth, enhancing the efficiency of large model training [3][23] - The supernode architecture is designed to integrate hundreds of processors, reshaping the competitive landscape in AI infrastructure [22][24] Summary by Sections 1. Huawei Computing Power Product Launch - Ascend chips will follow a yearly iteration schedule, with the 910C launching in Q1 2025 and subsequent models (950PR, 950DT, 960, 970) planned through 2028 [14][15] - The Ascend 950 series introduces low-precision data formats and self-developed HBM, enhancing training efficiency for large models [15][16] - The Ascend 960 and 970 are expected to double performance metrics across various parameters, including computing power and memory bandwidth [18][19] 2. Supernode Products - The supernode data center (Atlas 900/950/960 SuperPoD) is designed for large-scale AI training, achieving EFLOPS-level computing power with high bandwidth and low latency [2][27] - The supernode cluster (Atlas 950/960 SuperCluster) enhances network performance and energy efficiency, reaching ZFLOPS-level computing power [2][37] - The enterprise-grade air-cooled supernode server (Atlas 850) is tailored for post-training and multi-scenario inference, supporting flexible scaling [2][38] 3. Related Investment Targets - Key investment targets include hardware partners for Ascend, domestic wafer foundries, and companies involved in copper connections, optical connections, power supplies, PCBs, and cooling solutions [4][46]

Industry Investment Rating - The investment rating for the computer industry is "Outperform the Market" and is maintained [1] Core Viewpoints - The report highlights the significant advancements in domestic chip capabilities, particularly through Huawei's release of the Ascend chip roadmap, which aims to double computing power annually to meet the growing demands of AI [4][5] - The report indicates that the self-sufficiency rate of domestic computing chips in China is expected to reach 82% by 2027, driven by policy guidance and performance improvements [7] - The report suggests focusing on the computing power supply chain, including various companies across different segments such as Huawei, Alibaba, and Baidu, which are leading the charge in domestic chip development [9] Summary by Relevant Sections Industry Basic Situation - The closing index for the computer industry is 5648.76, with a 52-week high of 5841.52 and a low of 2855.49 [1] Recent Developments - Huawei plans to launch several new Ascend chip products between 2026 and 2028, enhancing AI computing capabilities [4][5] - Alibaba's AI computing card has been showcased as outperforming NVIDIA's A800 in several hardware parameters [6] - Baidu's Kunlun chip has secured significant orders, indicating strong market demand for domestic AI computing solutions [6] Investment Recommendations - The report recommends monitoring various supply chains related to computing power, including companies like Huawei, Alibaba, and others across different technology segments [9]