Core Viewpoint - Shenzhen has released the "Artificial Intelligence+" Advanced Manufacturing Action Plan (2026-2027), aiming to integrate AI technology deeply into the manufacturing process and enhance the semiconductor industry through AI applications [1][3]. Group 1: Key Support Platforms - The plan emphasizes the establishment of an Industrial Intelligence Innovation Center and aims to accelerate the construction of provincial and national manufacturing innovation centers [5]. - It supports the development of industrial software and knowledge alliances, focusing on standardizing industrial knowledge and creating leading autonomous industrial software products [5][6]. Group 2: Empowering Key Industry Clusters - AI will be applied to critical areas of the semiconductor industry, optimizing chip design and software coding efficiency, with a focus on developing high-performance SoC chips for various AI applications [8]. - The plan also highlights the importance of AI in robotics, promoting the development of multimodal interaction technologies and enhancing the capabilities of robots in industrial settings [8][9]. - In the low-altitude economy, the plan aims to establish a drone autonomy evolution system and enhance drone capabilities for various applications, improving resource scheduling and operational efficiency [9]. Group 3: AI in Electronics and Healthcare - The plan encourages collaboration among leading enterprises in the electronic information manufacturing sector to explore potential AI application scenarios and develop benchmark projects [9]. - In the pharmaceutical and medical device sectors, the plan promotes the integration of AI in drug development and medical technology, aiming to accelerate innovation and enhance the capabilities of medical devices [10].

中化新网讯日前，中国信息通信研究院(以下简称中国信通院)正式启动首批可信AI工业智能体评估工 作。 该评估依据中国信通院人工智能研究所依托中国人工智能产业发展联盟智能体创新与应用工作组，联合 上海移动、南方电网、广西电网、中国石化、中国石油等多家企业共同编制完成的《智能体技术要求与 评估方法行业应用工业》技术规范，立足工业行业对复杂性、可靠性的要求，结合智能体的能力特性， 从技术到应用展开全面评估，覆盖基础能力、业务场景、服务应用3大能力域，共计20余个能力项。 其中，基础能力部分主要评估工业智能体在感知、认知、决策、执行等方面的基本技术能力，包含工业 数据采集、工业数据加工、机理融合、生产规划、协同控制等能力项；业务场景主要评估工业智能体场 景应用的丰富程度，包含研发设计、工艺仿真等产品研发场景，生产优化、运行维护、质量控制等生产 管理场景，供应链管理、经营管理等运营管理场景；服务应用主要评估工业智能体服务应用的成熟度， 包含业务效果、智能交互、混合部署、系统兼容、安全保障、运维监控等能力项。(吴班) ...

一份刚刚发布的三年行动方案，将重塑这座城市的产业版图，并将触角伸向近地轨道。 2025年，上海规上工业总产值跨越4万亿元大关，创下历史新高。在这个制造业强势复苏的基础上，上 海并未停下脚步。近日，上海市经济和信息化委员会牵头编制的《上海市支持先进制造业转型升级三年 行动方案（2026—2028年）》（下称《行动方案》）正式发布，设定了明确目标：到2028年，新增年产 值10亿元以上制造业企业100家，累计超过600家，并带动产业链新增规上工业企业500家。这个被称 为"10亿+"的企业集群，是上海工业经济的核心力量。 01 产业投资的强劲势头 强劲的投资为未来增长注入了信心。数据显示，2025年，上海工业投资同比增长20.0%。其中，作为核 心的制造业投资表现尤为亮眼，同比增长22.8%。这不仅体现了市场主体的预期，也为《行动方案》提 出的三年目标提供了坚实基础。 02 多路径的产业结构调优 《行动方案》提出了以四大行动、17条举措为核心的完整路径。其中，"结构调优升级行动"对不同能级 的产业给出了差异化方案。对于传统优势产业，方案强调"优化提升"。例如，推动石化企业向"去油增 化"转型，布局新型功能材料；钢 ...

Core Viewpoint - The article discusses the implementation of the "AI + Manufacturing" initiative by various Chinese government departments, aiming to integrate artificial intelligence deeply into the manufacturing sector by 2027, with specific targets and deliverables outlined [2][3]. Group 1: Implementation Goals - By 2027, the initiative aims to apply 3-5 general large models in manufacturing, produce 1,000 high-level industrial intelligent entities, create 100 high-quality industrial data sets, and promote 500 typical application scenarios [3]. - The initiative emphasizes the development of 2-3 ecosystem-leading enterprises and the identification of 1,000 benchmark enterprises [3]. Group 2: Technical Focus - The implementation plan shifts focus from "smart factories" to a framework of "computing power - models - data - scenarios," highlighting the need for enhanced computing power supply and support for key technologies such as training chips and AI servers [4]. - It also calls for the cultivation of industry-specific large models and the development of a "cloud-edge-end" model system to facilitate lightweight deployment in industrial scenarios [5]. Group 3: Specific Industry Impact - For the lithium battery industry, the document indicates a shift from isolated algorithms to replicable capabilities at the production line level, emphasizing the importance of data loops in maintaining quality, safety, and consistency [6]. - The initiative specifies that large models should be deeply embedded in core manufacturing processes, targeting areas such as production scheduling, process optimization, and predictive maintenance [6]. Group 4: Safety and Compliance - The document outlines the need for advancements in key technologies related to industrial model safety, including deep synthesis verification, algorithm security, and training data protection [7]. - It emphasizes the establishment of safety policies and standards for AI in the industrial sector, particularly for high-safety-sensitive industries like lithium batteries [8][9]. Group 5: Funding and Support - The initiative proposes coordinated funding channels to support "AI + manufacturing" technology research and application, leveraging national AI industry investment funds [9][10]. - Local governments are encouraged to provide support through "computing power vouchers" and "model vouchers" to guide development and prevent excessive competition [10].

工业AI 的需求已从早期头部企业的探索性投入，转向全行业"提质降本增效"的刚需。根据IDC预测，到2028年，中国工业企业AI 支出规模将接近90 亿元 人民币，年复合增长率达到 38%。 到2030年，全球活跃智能体数量将突破 22.16 亿个，年复合增长率达到 139%，其中工业领域的活跃智能体是最重要组成部分之一。IDC 认为，智能体数 量的快速增长将与工业 AI 的规模化需求形成共振：一方面，工业企业对跨系统协同与流程闭环的诉求更强；另一方面，智能体作为"任务编排与流程执行 载体"，有助于将 AI 从"点状能力"升级为"可运营的生产力"，从而加速规模化落地。 智通财经APP获悉，国际数据公司IDC发文称，工业AI 的需求已从早期头部企业的探索性投入，转向全行业"提质降本增效"的刚需。根据IDC预测，到 2028年，中国工业企业AI 支出规模将接近90 亿元人民币，年复合增长率达到38%。到2030年，全球活跃智能体数量将突破 22.16 亿个，年复合增长率达 到 139%，其中工业领域的活跃智能体是最重要组成部分之一。 当前，中国工业行业正处于由"数字化"向"智能化"跨越的关键拐点，"人工智能+工业" ...

Group 1 - The integration of AI large models is driving new modes such as platform design, intelligent production, and personalized customization, while open-source AI technologies are lowering the barriers for enterprises to adopt intelligent solutions [1] - For large enterprises, the focus should be on leveraging industrial software to apply AI technologies, while for small and medium-sized enterprises, a "Model as a Service (MaaS)" approach is recommended to reduce the technical challenges of AI adoption [1] - The upgrade of industrial networks is crucial for the deep integration of AI and industry, with a policy goal to promote the transformation of at least 50,000 enterprises by 2028 [1] Group 2 - Future efforts should focus on innovating new industrial networks, building high-quality industrial data sets, and cultivating industrial intelligence, emphasizing real-time, deterministic, and secure network capabilities [2] - The establishment of innovation coalitions in key industries such as automotive and steel is necessary to create valuable industrial data sets and develop a multi-level application system [2] - The development of industrial intelligence should follow a step-by-step approach from device level to enterprise level, enhancing capabilities progressively [2]

Core Viewpoint - The company, Koyuan Smart (002380.SZ), anticipates a significant decline in net profit for 2025, with projections indicating a decrease of 48.38% to 32.49% compared to the previous year, while also reporting a robust growth in its core business driven by industrial automation and AI strategies [1][2]. Financial Performance - The company expects net profit attributable to shareholders for 2025 to be between 130 million to 170 million yuan, a decline of 48.38% to 32.49% year-on-year [1]. - The net profit after excluding non-recurring gains and losses is projected to be between 100 million to 140 million yuan, reflecting a decrease of 56.84% to 39.58% compared to the previous year [1]. - For the reporting period, the company achieved an operating revenue of approximately 1.86 billion yuan, representing a year-on-year growth of 10.90% [2]. - After adjusting for impairment factors related to biomass energy assets, the net profit attributable to the parent company is expected to be around 350 million yuan, marking a year-on-year increase of 39.57% [2]. Strategic Focus - The company is firmly focused on two strategic directions: industrial automation and industrial artificial intelligence, continuing to deepen its engagement in foundational industries such as energy, metallurgy, chemicals, building materials, and water services [2]. - The company is experiencing a significant release of demand for autonomous control in the industrial sector, with a notable increase in clients' willingness to upgrade to digital and intelligent solutions [2]. Business Development - The core business operations are reported to be stable, with an optimized business structure and sustained resilience in development [2]. - The company has established a comprehensive "Industrial Intelligence" system that includes applications such as intelligent inspection, monitoring, unattended operations, and smart decision-making, facilitating the transition from "human-operated" to "unmanned" industrial sites [2]. - Industrial AI-related business revenue reached approximately 260 million yuan, accounting for over 10% of the company's total operating revenue for the first time, indicating a shift from the "layout phase" to the "expansion phase" of the second growth curve [2].

（来源：多知） 多知1月19日消息，工信部等八部门近日联合印发《"人工智能+制造"专项行动实施意见》，系统部署以人工智能技术推动制造业智能化升级。 在总体要求上，文件提出： 到2027年，我国人工智能关键核心技术将实现安全可靠供给，产业规模和赋能水平稳居世界前列。 推动3-5个通用大模型在制造业深度应用、推出1000个高水平工业智能体、打造100个工业领域高质量数据集、推广500个典型应用场景。 培育2-3家具有全球影响力的生态主导型企业和一批专精特新中小企业，打造一批"懂智能、熟行业"的赋能应用服务商，选树1000家标杆企业。 建成全球领先的开源开放生态，安全治理能力全面提升，为人工智能发展贡献中国方案。 围绕教育科技与人才培养，文件中也多有涉及。 在加强人才引育方面，文件提到： 开展人工智能产业人才需求预测，发布人才需求预测报告，支持高校院所提前布局、调整优化相关学科专业。 建好用好北京中关村学院、上海创智学院、深圳河套学院、国家人工智能产教融合创新平台、国家卓越工程师学院、国家卓越工程师实践基地等，设置专 业课程，培养既懂人工智能又懂制造业应用的复合型人才，完善人工智能认知教育培训，提升全员人工智能素养 ...

2026.01. 15 本文字数：1597，阅读时长大约3分钟 金江对第一财经记者表示，此次对外发布的40多个工业智能体，是从美的内部上万个智能体中筛选 而来，目前美的已经在158个核心业务场景中实现规模化应用，覆盖研发、制造、供应链、品质、物 流、财务、人力资源和海外服务等环节。 过去两年，AI在企业内部最先落地的是翻译、文档、人力资源等通用场景，但对成本结构和生产效 率影响更大的应用，正在转向对制造流程的深度嵌入。 "现在的重点已经不是做多少个智能体，而是哪些场景真的能带来效率和成本改善。"金江说。 智能体走进生产系统 在制造环节，美的已经将智能体用于多类具体生产管理场景。 例如，在产线管理中，美的部署了用于损失分析的TPM、DMS、LOSS等智能体。过去，产线因停 机、质量问题等造成的损失通常需要人工每隔数小时盘点一次，很难迅速定位原因；现在通过智能 体，可以将每天的损失拆解到具体工序和岗位，并给出针对性的改善建议。 作者 | 第一财经 王珍 封图 | 由受访者提供 在经历了两年对大模型的试验性部署后，中国制造业正把人工智能的应用重点，从通用办公场景推向 生产一线。 1月15日，美的集团（000333. ...

金江对第一财经记者表示，此次对外发布的40多个工业智能体，是从美的内部上万个智能体中筛选而来，目前美的已经在158个核心业务场景中实现规模化 应用，覆盖研发、制造、供应链、品质、物流、财务、人力资源和海外服务等环节。 过去两年，AI在企业内部最先落地的是翻译、文档、人力资源等通用场景，但对成本结构和生产效率影响更大的应用，正在转向对制造流程的深度嵌入。 "现在的重点已经不是做多少个智能体，而是哪些场景真的能带来效率和成本改善。"金江说。 中小企业应用工业智能体将会面临新的挑战，一是应用效果依赖基础数据，二是要有AI人才，三是还要找对应用场景。 在经历了两年对大模型的试验性部署后，中国制造业正把人工智能的应用重点，从通用办公场景推向生产一线。 1月15日，美的集团（000333.SZ）旗下数字化子公司美云智数发布工业智能体矩阵，并同步推出美擎AIGC 3.1平台。美云智数总裁金江在发布会上表示，美 的计划在2026年前通过AI相关应用实现约9亿元的成本节约。 智能体走进生产系统 在制造环节，美的已经将智能体用于多类具体生产管理场景。 海尔集团旗下的卡奥斯工业互联网平台表示，其工业大模型已在石油化工、能源、家电等 ...