Core Viewpoint - Moores Threads has launched its new GPU architecture "Huagang" at the MUSA Developer Conference, showcasing significant advancements in computing power and energy efficiency [1][2]. Group 1: Product Launch and Features - The "Huagang" architecture features a 50% increase in computing density and a 10-fold improvement in energy efficiency, supporting large-scale intelligent computing clusters of over 100,000 cards [1][2]. - The full-featured GPU includes four main functional engines: AI computing acceleration, graphics rendering, physical simulation and scientific computing, and ultra-high-definition video encoding and decoding [1]. - The company plans to release high-performance AI training and inference chip "Huashan" and a chip focused on high-performance graphics rendering "Lushan" based on the new architecture [2]. Group 2: Market Position and Financial Performance - Moores Threads is regarded as the "Chinese version of Nvidia" and recently went public, experiencing a significant stock price increase of over 400% on its first trading day [6]. - The stock price has seen fluctuations, currently at 664.1 yuan per share, down from a peak of 940 yuan [6]. - For the first nine months of 2025, the company reported a revenue of 785 million yuan and a net loss of 724 million yuan, with projections indicating a continued net loss for the year [7].

12月18日，在2025光合组织人工智能创新大会（HAIC 2025）现场，广州国家实验室、天津大学、湖南 应用数学中心、中科院高能所、国家天文台、中科院大气所、中石油东方物探、中科曙光、合肥大数据 公司等22家高校、科研机构及企业共同发起 "科学智能联合攻关行动" 。该行动将重点围绕科学大模型 开发、超智融合算力平台建设、模型训练推理优化、科学数据开放共享等方面开展协同工作。光合组织 透露，在未来三年内预计将推动超过20个行业级AI4S软硬协同解决方案落地。 ...

来源：北京科协 你是否曾仰望星空，思考宇宙的边界？ 你是否好奇，构成这个世界的最小单元，到底是什么？ 当人工智能闯入科学的世界，一切认知，正在被重新书写。 带上你的好奇心，让我们一起走进这场正在发生的科技革命。 科学，不再只是试管、望远镜与公式推演； AI，正成为我们理解宇宙、探索未知的"新器官"。 从宏观层面—— 太阳耀斑能预测吗？怎么提高准确率？ 到微观世界—— 如何在海量数据中找到极少数信号粒子出现的事例？ 从"人工操控"—— 人类做不到的精雕细琢，人工智能模型能完成吗？ 到"智能自驱"—— AI设计的火箭发动机真的能用吗？ 2025年7月，北京发布全国首个专注于"AI for Science"的专项地方政策，推动人工智能与科研的深度融 合。这不是技术的简单叠加，而是一场科研范式的根本变革。 （来源：北京科普发展与研究中心） 北京正成为科学智能的"策源地" 让我们一起见证 科学智能的黄金时代 "科技创新调研行"——六集"人工智能+"主题纪录片《AI向新力》，由北京市科学技术协会出品，北京 科普发展与研究中心承制，通过纪实镜头全景展现首都以人工智能为核心引擎，培育和发展新质生产力 的生动实践。 敬请关注" ...

今年8月，国务院印发《关于深入实施"人工智能+"行动的意见》为我国推动人工智能与经济社会各行业 各领域广泛深度融合提供了指引。10月28日发布的《中共中央关于制定国民经济和社会发展第十五个五 年规划的建议》中再次明确："深入推进数字中国建设""加快人工智能等数智技术创新""全面实施'人工 智能+'行动，以人工智能引领科研范式变革，加强人工智能同产业发展、文化建设、民生保障、社会治 理相结合，抢占人工智能产业应用制高点，全方位赋能千行百业"。 从线上人工智能（以下简称"AI"）大模型与教育、医疗、金融等各行各业深度绑定，持续刷新行业效率 上限；到线下具身智能机器人在工厂协作、社区养老、家庭服务中崭露头角，为人类生产生活带来无限 可能……2025年，人工智能正以前所未有的速度穿透虚拟与现实、串联技术与产业，也让人们不禁畅 想，AI技术的未来发展还能带来怎样的惊喜。 不过，在AI技术打破人机边界的背后，安全治理也成为不可忽视的命题。姚期智提醒，AI算法潜在的 不可靠性可能引发隐私泄露、冲击社会价值伦理等风险。"目前中外正在探索将AI与密码学、博弈学等 理论结合的交叉领域，凝聚国际共识，携手构建AI治理协议。"姚期智 ...

Core Insights - Google DeepMind's AlphaFold has significantly impacted protein structure prediction, driving advancements in scientific research over the past five years [1][4] - AI is reshaping scientific research, particularly in life sciences and biomedicine, due to rich data availability and urgent societal needs [1][3] Group 1: AI in Scientific Research - AI models and tools have achieved breakthroughs in basic research, including protein structure prediction and the discovery of new biological pathways [1][3] - The paradigm of "foundation models + research agents + autonomous laboratories" is emerging in AI-driven scientific research [3][13] Group 2: Advancements in Biology - DeepMind's AlphaFold has solved the protein structure prediction problem, earning the 2024 Nobel Prize in Chemistry and establishing itself as a digital infrastructure for modern biology [4] - The C2S-Scale model, developed by Google and Yale University, has generated new hypotheses about cancer cell behavior, showcasing AI's potential in formulating original scientific hypotheses [8] Group 3: AI in Drug Development - AI-assisted pathology detection has expanded to new disease scenarios, with the DeepGEM model achieving a prediction accuracy of 78% to 99% for lung cancer gene mutations [10] - The AI-optimized drug MTS-004 has completed Phase III clinical trials, marking a significant milestone in AI-driven drug discovery [10] Group 4: AI in Other Scientific Fields - AI applications in materials science are gaining momentum, with startups like Periodic Labs and CuspAI focusing on discovering new materials [11] - DeepMind's WeatherNext 2 model has surpassed traditional physical models in accuracy and efficiency for weather predictions [5] Group 5: Future of AI in Science - The evolution of scientific intelligence technologies is expected to accelerate, with AI foundational models and robotics enhancing research efficiency [19] - The integration of AI into scientific discovery is anticipated to lead to significant breakthroughs, with predictions of achieving near-relativistic level discoveries by 2028 [19]

Investment Rating - The report does not explicitly provide an investment rating for the industry. Core Insights - The integration of artificial intelligence (AI) with energy systems is identified as a crucial driver for high-quality energy development, aiming to enhance the reliability and efficiency of energy systems while supporting green and low-carbon transitions [5][7][10]. - The report emphasizes the importance of data-mechanism fusion technology in advancing AI applications within the energy sector, particularly in the context of new power distribution systems [18][32]. - The future outlook includes the development of a digital twin system for new power distribution systems, which will enhance decision-making and risk management through real-time data integration and intelligent feedback mechanisms [111][113]. Summary by Sections 1. Background Significance - The report highlights the role of AI as a significant driver of productivity and a key component in the new energy landscape, particularly in the context of China's dual carbon goals [5][8][12]. - It discusses the challenges faced by traditional AI methods in energy systems, such as sample dependency and limited generalization capabilities, which necessitate the adoption of data-mechanism fusion approaches [15][18]. 2. Key Technologies - **Scientific Intelligence**: Defined as a new paradigm in scientific research driven by AI, focusing on learning, simulation, prediction, and optimization to facilitate scientific discovery [23][25]. - **Data-Mechanism Fusion**: This method combines data-driven and mechanism-driven approaches to enhance model accuracy and decision-making reliability in energy systems [32][44]. - The report outlines five fusion models: serial, feedback, parallel, guiding, and embedding, which facilitate the integration of data and mechanism knowledge [48][51]. 3. Application Exploration - **Calculation Inference**: The report details a two-phase method for identifying line parameters in distribution networks, which significantly reduces errors in parameter estimation [66][73]. - **Source-Load Forecasting**: It discusses the development of a core technology system for high-precision forecasting of source-load dynamics, addressing challenges such as complex temporal data relationships and information gaps [76][78]. - **Operational Optimization**: The report presents a framework for edge-cloud collaborative optimization, enhancing computational efficiency and data privacy in regional energy internet applications [87][89]. 4. Future Outlook - The report envisions a future where scientific intelligence and data-mechanism fusion technologies will drive innovations in power system operations, scheduling, and decision-making [104][107][109]. - It emphasizes the need for robust and interpretable models that can adapt to dynamic and uncertain environments, ensuring the safety and reliability of power grid operations [113].

Core Viewpoint - The rapid rise of "AI for Science" is a significant topic, emphasizing the need for researchers at the forefront of science to integrate AI with traditional technologies, as methods of scientific research will evolve dramatically in the coming years [1] Group 1 - The "2025 AI+" conference highlights the importance of adapting to the integration of AI in scientific research [1] - Yao Qizhi, a Turing Award winner and academician, stresses the urgency for scientists to understand and leverage the development trends of AI [1] - The methods of conducting scientific research are expected to change significantly within the next five to ten years [1]

Core Insights - The 2025 Xili Lake Forum focuses on accelerating scientific discovery and defining the future of industries, emphasizing the integration of artificial intelligence and scientific research [1] - Key announcements include the establishment of the International Intellectual Property Academy and the launch of the Peking University AI4S Talent Cultivation Program [1] Group 1: Forum Highlights - The forum gathers academic leaders, industry pioneers, and cross-disciplinary forces to discuss topics related to scientific infrastructure in the AI era and seizing opportunities in new industrial transformations [1] - Zhang Jin, a prominent academic figure, highlights "scientific intelligence" as a crucial innovation merging AI with scientific research, marking it as a significant outcome of the new technological revolution [1] Group 2: Strategic Importance - The forum underscores Shenzhen's commitment to enhancing original innovation and investing in talent, which is deemed significant in the context of global industrial restructuring [2] - Lin Ziao, CEO of Aosheng Cheng, emphasizes that accelerating scientific discovery will grant greater influence in future industries [2]

Core Insights - The forum "AI for Science" held during the 2025 World Internet Conference focused on how AI is reshaping scientific research paradigms and stimulating new productivity [2][3][4] Group 1: AI Applications in Scientific Research - AI is becoming a crucial tool to overcome long-standing challenges in materials research, such as measurement limitations, as highlighted by Chen Lidong from the Shanghai Institute of Silicate [3] - AI models have shown significant potential in enhancing catalyst performance by 50% through iterative experimentation and modeling, demonstrating the efficiency of AI in material discovery [3] - The concept of "AI for Materials" and "Materials for AI" emphasizes a reciprocal relationship between AI and materials science [4] Group 2: AI in Healthcare - AI brain-machine interfaces are being applied in managing neurodegenerative diseases, with advancements allowing for quicker detection of seizures compared to traditional methods [5] - The accuracy of language decoding in AI has improved significantly, particularly in recognizing Chinese phonetics, achieving over 70% accuracy [5] Group 3: AI's Impact on Innovation - Generative AI is optimizing product design and team collaboration in open innovation, while its direct impact on disruptive innovation remains limited, underscoring the importance of human creativity [7] Group 4: Future Directions in AI and Science - The "scientific intelligence" concept is seen as a pathway to superintelligence, with significant advancements in drug design for diseases lacking clear targets, achieving a 50-fold improvement in molecular design [9][10] - The demand for computational power in AI for science is growing exponentially, necessitating a unique capability to couple high and low precision in scientific calculations [11] - The release of the "Global AI Standards Development Report" calls for collaboration among international organizations, governments, and industries to establish responsible global standards [13] Group 5: AI Infrastructure and Talent Development - The "Panshi V1.5" platform aims to empower scientific research across disciplines, covering the entire research process from hypothesis to discovery [18] - The forum concluded with discussions on the role of AI in empowering young scientists, emphasizing the need for interdisciplinary collaboration and the evolution of AI from a tool to a collaborator [25]

Core Insights - The "Panshi V1.5: One-stop Research Platform" was officially launched at the 2025 World Internet Conference, marking a significant evolution from the previous version released on July 26 this year [1] - The V1.5 upgrade enhances the foundational capabilities of "Panshi: Scientific Foundation Model" and "Panshi: Literature Compass," while introducing two new scientific intelligent agents: "Panshi: Innovation Assessment" and "Panshi: Intelligent Factory," making the platform more comprehensive [1] Group 1: Applications in Astrophysics - In the field of astrophysics, a star parameter inversion toolchain was developed in collaboration with the National Astronomical Observatory, addressing high computational costs and complex processes [2] - The new system transforms traditional complex numerical calculations into efficient interpolation and weighted matching, significantly improving inversion speed and enhancing result reliability, stability, and interpretability [2] - This advancement reduces computational costs and lowers the usage threshold, enabling cross-disciplinary researchers to easily conduct star parameter analysis and validation [2] Group 2: Innovations in Energy Materials - In the energy materials sector, a fully automated end-to-end material reverse design system, S1-MatAgent, was created in partnership with the Shanghai Institute of Ceramics, Chinese Academy of Sciences [2] - This system autonomously performs literature reading, material calculations, and optimizations, successfully identifying 13 high-performance materials from 20 million candidate formulations, with new materials showing a 38% improvement in activity over traditional commercial catalysts [2] - The design cycle, which previously took months, has been reduced to just 30 minutes, marking a critical shift from "trial-and-error" to "AI-driven" material development [2] Group 3: Advancements in Mechanical Engineering - In mechanical engineering, an intelligent load calculation technology was developed in collaboration with the Institute of Mechanics, Chinese Academy of Sciences, addressing high costs and long cycles in fluid load calculations for complex configurations like high-speed trains and aircraft [3] - This technology reduces key parameter errors by 42% in data-scarce scenarios and shortens the simulation analysis time for high-speed train aerodynamic issues from several hours to seconds [3] - It supports multi-format 3D configuration input and automates the entire process from data parsing to result visualization, providing critical data support for the design and optimization of major equipment configurations [3]