机构：中金公司 研究员：张怡康/彭虎/贾顺鹤 4Q25 业绩符合我们预期 公司发布4Q25 业绩，H 股口径收入6.6 亿美元，YoY+22.4%，QoQ+3.9%，得益于晶圆出货量增加和平 均售价提升。毛利率为13%，YoY+1.6ppt，得益于ASP 改善和成本削减；但因劳动力成本上升，QoQ- 0.5ppt。归母净利润0.17 亿美元，YoY 扭亏为盈，QoQ 小幅下滑。公司指引1Q26 收入6.5 亿~6.6 亿美 元，毛利率预计13%~15%。4Q25 业绩、1Q26 指引符合我们预期。 发展趋势 4Q25 营收分析：4Q25 嵌入式收入YoY+31.3%；独立非易失性存储器收入YoY+22.9%，功率收入 YoY+2.4%；射频逻辑收入YoY+19.2%；模拟与电源管理收入YoY+40.7%。总的来看，MCU、智能芯 片、闪存、电源管理芯片需求较强，功率、消费类产品增长平稳，从下游分布表现为4Q25 公司计算类 产品收入YoY+69.9%，通信收入YoY+29.1%。因区域性的订单变化和在地生产趋势，2025 年中国地区 营收YoY+19.6%；北美地区营收YoY+51.3%；欧洲地区营收YoY+ ...

（文章来源：证券日报） 证券日报网讯 2月12日，寒武纪在互动平台回答投资者提问时表示，在大模型技术革新的背景下，智能 计算需求继续增长。公司凭借在智能芯片领域里具有竞争力的产品优势及良好的服务，获得了客户的广 泛认可，在市场中赢得了良好的口碑。公司将持续密切关注行业发展和市场需求，把握算力机遇，加深 与客户侧的合作，持续聚焦人工智能芯片设计领域技术创新，提升核心竞争力，积极拓展市场份额，加 速场景落地，推动公司的持续良好发展。 ...

Core Viewpoint - The industrial internet, as a key engine for industrial transformation and competitive advantage, is advancing in China with the establishment of a unique identification and resolution system that enables data interoperability and value sharing across various sectors and regions [1][2]. Group 1: Development of the Identification and Resolution System - China's identification and resolution system for the industrial internet has transitioned from a "construction period" to an "application period" and "integration period," marking a significant advancement in its capabilities [1]. - The system has achieved controllable architecture, facilities, and technology, becoming an essential infrastructure for interconnectivity among various entities since its initiation in 2018 [1][2]. - The release of the "2024-2026 Action Plan" by the Ministry of Industry and Information Technology and other departments signifies the start of a new phase focused on large-scale applications [1]. Group 2: Trends and Applications - The recently published "Case Collection" includes 24 benchmark cases across 19 key industries and 5 major fields, showcasing the current state of industry development [2]. - Three major trends in the industry are identified: moving from internal management to full-process collaboration, from single-function applications to integrated innovation, and from key industries to broader fields [2]. - The application of identification technology has expanded from internal production tracking to encompassing the entire value chain, including R&D, production, supply chain, marketing, service, and recycling [2]. Group 3: Integration with New Technologies - Identification resolution is evolving beyond a mere tracking tool, integrating with technologies like 5G, artificial intelligence, and digital twins to create innovative applications [3]. - For instance, in the photovoltaic industry, a distributed certification platform combining identification and blockchain technology has been developed to provide each device with an immutable "digital identity," supporting green electricity trading and carbon footprint tracking [3]. Group 4: Value Creation and Impact - The application of identification resolution technology has led to significant improvements in operational efficiency and competitiveness for various companies [3]. - For example, a company increased its custom product order capacity from 5% to 30%, reduced changeover costs by 70%, and shortened order delivery cycles by 40% [3]. - The integration of the identification resolution system has improved on-time delivery rates to 98.5%, enhanced overall production efficiency by approximately 10%, and reduced inventory turnover from 226 days to 171 days [3].

Core Insights - The article discusses the implementation of the "Artificial Intelligence + Manufacturing" initiative by eight departments, aiming to accelerate the integration of AI with the manufacturing sector and promote intelligent transformation [2][4] - The Ministry of Industry and Information Technology (MIIT) emphasizes that developing AI is essential for achieving new industrialization, focusing on both technology supply and application empowerment [3] - The initiative outlines seven key tasks and 21 specific measures to enhance the intelligent, green, and integrated development of the manufacturing industry [4][5] Group 1: Key Tasks and Goals - The initiative aims to apply 3 to 5 general large models in manufacturing by 2027, create 100 high-quality industrial data sets, and promote 500 typical application scenarios [4] - It seeks to cultivate 2 to 3 globally influential leading enterprises and a number of specialized small and medium-sized enterprises, establishing a robust ecosystem for AI in manufacturing [4] - The goal is to build a globally leading open-source ecosystem and enhance security governance capabilities [4] Group 2: Industry-Specific Guidelines - The "Guidelines for Key Industry Transformation" provide direction for sectors such as raw materials, equipment manufacturing, consumer goods, electronics, and software services [6] - In the automotive sector, the guidelines emphasize accelerating the intelligent upgrade of the entire supply chain, including the development of automotive large models for design optimization [6] - The integration of AI in manufacturing is seen as a critical factor for driving innovation in research and design, improving production efficiency, and optimizing supply chain management [6][11] Group 3: Industrial Internet and AI Integration - The integration of industrial internet and AI is highlighted as a key driver for new industrialization, with plans to enhance the level of integration by 2028 [7][8] - The initiative includes expanding new industrial networks to meet the demands of AI applications and improving data governance and sharing systems [7] - The focus is on creating a collaborative ecosystem that promotes the application of AI and industrial internet across various key industries [7][8] Group 4: Local Initiatives and Developments - Various local governments are launching their own "AI+" initiatives to drive high-quality development, with specific action plans and targets [9][10] - For instance, Shanghai aims to add 100 manufacturing enterprises with over 1 billion yuan in annual output by 2028 [9] - The establishment of dedicated AI development agencies in cities like Guangzhou and Zhuhai reflects a growing commitment to fostering AI innovation at the local level [10]

Core Viewpoint - The article discusses the implementation of the "Artificial Intelligence + Manufacturing" initiative by the Ministry of Industry and Information Technology (MIIT) and other departments, aiming to accelerate the integration of AI with the manufacturing sector and promote intelligent transformation in the industry [3]. Group 1: Key Tasks and Goals - The initiative outlines seven key tasks including innovation foundation, intelligent upgrade, product breakthroughs, entity cultivation, ecological expansion, safety protection, and international cooperation, with 21 specific measures to enhance the intelligent, green, and integrated development of manufacturing [5]. - By 2027, the goal is to promote the deep application of 3 to 5 general large models in manufacturing, create 100 high-quality industrial data sets, and establish 500 typical application scenarios [5]. Group 2: Industry Transformation Guidelines - The "Guidelines for Key Industry Transformation Empowered by AI" provide direction for sectors such as raw materials, equipment manufacturing, consumer goods, electronic information, and software services [6]. - In the automotive sector, the guidelines emphasize accelerating the intelligent upgrade of the entire supply chain, including the development of automotive large models for design and optimization [6]. Group 3: Industrial Internet and AI Integration - The integration of industrial internet and AI is highlighted as a crucial aspect of the initiative, with plans to enhance the level of integration by 2028, including the deployment of new industrial networks in over 50,000 enterprises [9]. - The article notes that AI is transitioning from being an auxiliary tool to becoming a core engine for industrial internet, enhancing capabilities such as real-time optimization and intelligent scheduling [10]. Group 4: Local Initiatives and Developments - Various local governments are actively promoting AI initiatives, with Shanghai aiming to add 100 manufacturing enterprises with over 1 billion yuan in annual output by 2028 [12]. - The establishment of dedicated AI development agencies in cities like Guangzhou and Zhuhai reflects a growing commitment to fostering AI innovation and integration within local industries [13].

Core Viewpoint - The Chinese government is accelerating the integration of artificial intelligence (AI) with the manufacturing sector to promote intelligent transformation and drive economic growth through the issuance of the "AI + Manufacturing" implementation opinions and the "Industrial Internet and AI Integration Empowerment Action Plan" [1][2]. Group 1: Key Tasks and Goals - The "Opinions" outline seven key tasks including innovation foundation, intelligence upgrade, product breakthroughs, entity cultivation, ecological expansion, safety protection, and international cooperation, with 21 specific measures to enhance the intelligent, green, and integrated development of manufacturing [2][3]. - By 2027, the plan aims to implement 3 to 5 general large models in manufacturing, create 100 high-quality industrial data sets, and promote 500 typical application scenarios [2][3]. Group 2: Industry Focus and Applications - The focus industries for AI integration include raw materials, equipment manufacturing, consumer goods, electronic information, and software services, with specific guidance provided for each sector [3][4]. - In the automotive sector, the plan emphasizes the intelligent upgrade of the entire supply chain, including the development of large models for automatic design and optimization of vehicle parameters [3][4]. Group 3: Infrastructure and Support - The "Action Plan" emphasizes the need for new industrial networks to support AI applications, targeting the deployment of these networks in over 50,000 enterprises by 2028 [6][7]. - The plan also aims to enhance the quality of industrial data collection and sharing in 20 key industries, establishing a robust data governance framework [6][7]. Group 4: Local Initiatives - Various local governments are launching their own AI initiatives, such as Shanghai's three-year action plan to support advanced manufacturing and Beijing's plan to establish itself as an AI innovation hub [8][9]. - The establishment of dedicated AI development agencies in regions like Guangzhou and Zhuhai signifies a growing commitment to fostering AI-driven economic growth [9].

Core Viewpoint - The "Artificial Intelligence + Manufacturing" initiative aims to deeply integrate AI technology with the manufacturing industry, enhancing innovation and application capabilities by 2027 [1] Group 1: Key Objectives - By 2027, China aims to achieve secure and reliable supply of key AI technologies, maintaining a leading position in industry scale and empowerment levels [1] - The initiative plans to promote the deep application of 3-5 general large models in manufacturing, launch 1,000 high-level industrial intelligent entities, and create 100 high-quality industrial datasets [1] - The goal includes cultivating 2-3 globally influential leading enterprises and a number of specialized small and medium-sized enterprises, while establishing a leading open-source ecosystem [1] Group 2: Major Tasks - The initiative outlines seven key tasks including innovation foundation, intelligence upgrade, product breakthroughs, entity cultivation, ecosystem expansion, safety assurance, and international cooperation [1] - Specific measures include enhancing AI computing power, supporting the development of intelligent chips, and creating high-performance algorithm models tailored for manufacturing [2] - The initiative emphasizes the need for deep integration of AI in production processes, encouraging leading enterprises to pioneer AI applications and promoting digital empowerment for SMEs [2] Group 3: Product Development - The initiative aims to drive the iteration of intelligent equipment, integrating AI technology into the development and manufacturing of major technical equipment like aircraft and ships [3] - It also focuses on upgrading smart terminals, fostering innovation in AI-enabled products such as smartphones and smart home devices, and establishing testing bases for humanoid robots [3] - Policies will support the prevention of "involution" competition in the industry, coordinate funding channels for AI-related R&D, and enhance investment in quality projects [3]

Core Viewpoint - The article emphasizes the importance of artificial intelligence (AI) in China's economic development, highlighting the government's commitment to advancing AI technologies and ensuring a robust industrial ecosystem by 2027 [2][3]. Group 1: Government Initiatives - The Ministry of Industry and Information Technology, along with seven other departments, has issued the "Implementation Opinions on the 'Artificial Intelligence + Manufacturing' Special Action," aiming for secure supply of key AI technologies and maintaining a leading industrial scale by 2027 [2]. - The "14th Five-Year Plan" explicitly calls for the comprehensive implementation of the "Artificial Intelligence +" initiative, showcasing the government's strategic focus on AI development [3]. Group 2: Industry Development - China's AI industry has made significant progress, with a developing industrial chain from foundational support to core technologies and industry applications, leading to the emergence of innovative enterprises and new business models [2]. - The listing of Zhiyun (02513.HK), recognized as the world's first "large model stock," on the Hong Kong Stock Exchange exemplifies the growth and potential of China's AI sector [2]. Group 3: Competitive Advantages - Tesla's CEO Elon Musk has expressed confidence in China's AI capabilities, particularly its electricity supply, predicting that by 2026, China's power generation could reach three times that of the U.S., supporting energy-intensive AI data centers [3]. - China possesses significant policy, market, and innovation advantages that are crucial for the rapid development of its AI industry [3]. Group 4: Challenges and Focus Areas - Despite advancements, China's AI industry faces challenges such as weak core technologies and limited application scenarios, necessitating improvements in key areas [2][4]. - The "Implementation Opinions" stress the need for secure and reliable supply of core technologies, particularly in foundational theories, methods, and tools, including breakthroughs in chip manufacturing and application scenarios [4]. Group 5: Strategic Importance of AI - AI is becoming a new engine for economic development, with its strategic value surpassing that of any other single technology, making it a focal point of international technological and economic competition [5]. - Mastering core AI technologies and achieving a leading industrial scale is essential for seizing the opportunities presented by the AI revolution [5].

Core Viewpoint - The article discusses the implementation of the "Artificial Intelligence + Manufacturing" initiative, aiming to enhance the integration of AI technologies in the manufacturing sector by 2027, establishing China as a global leader in AI technology and application [1][6]. Group 1: Overall Requirements - By 2027, China aims to achieve secure and reliable supply of key AI technologies, maintaining a leading position in industry scale and empowerment levels globally. The initiative includes the application of 3-5 general large models in manufacturing, the launch of 1,000 high-level industrial intelligent entities, and the creation of 100 high-quality industrial data sets [6][5]. Group 2: Innovation Foundation - The initiative emphasizes strengthening AI computing power supply, supporting the development of intelligent chips and high-level computing facilities, and promoting the establishment of a national integrated computing network [7]. - It also focuses on developing high-performance algorithm models tailored for the manufacturing sector, encouraging the creation of industry-specific models and enhancing data governance capabilities [8][7]. Group 3: Application Empowerment - The initiative promotes the deep integration of AI in key manufacturing sectors, facilitating the development of intelligent factories and enhancing capabilities in design, production, and management processes [9][10]. - It encourages the establishment of AI application platforms to match supply and demand accurately, and supports the digital transformation of small and medium-sized enterprises [11][12]. Group 4: Product Breakthrough - The initiative aims to accelerate the iteration of intelligent equipment and the upgrade of smart terminals, including industrial robots and AI-driven medical devices, to enhance decision-making and operational capabilities [13][14]. - It also focuses on fostering new business models around intelligent entities and promoting the integration of AI with traditional software products [15][16]. Group 5: Ecological Growth - The initiative emphasizes the importance of standardization in AI development, encouraging participation in international standardization efforts and promoting an open-source ecosystem for AI projects [17][18]. - It aims to strengthen talent cultivation in the AI sector, ensuring a skilled workforce that understands both AI and manufacturing applications [19][18]. Group 6: Safety Assurance - The initiative outlines measures to enhance safety and security in AI applications, including the development of safety governance mechanisms and risk monitoring systems [20][21]. Group 7: International Cooperation - The initiative encourages international collaboration in AI, supporting enterprises in customizing AI products for different markets and participating in global AI discussions [21][20]. Group 8: Support Measures - The initiative calls for a comprehensive policy support system to facilitate the development of AI in manufacturing, including funding and resource allocation to promote new technologies and applications [22][23].

Core Insights - The semiconductor ETF Tianhong (159310) has achieved three consecutive days of gains, with a trading volume of 24.21 million yuan, reflecting a strong performance of the underlying index, the CSI Semiconductor Industry Index (H30007), which rose by 1.88% [1] - The Tianhong Science and Technology Innovation Index ETF (589860) also recorded three consecutive days of gains, with a trading volume of 35.92 million yuan, and the underlying index, the SSE Science and Technology Innovation Board Composite Index (000680), increased by 1.53% [1][2] - Samsung Electronics reported a remarkable 208% increase in fourth-quarter profit, driven by a surge in demand for AI servers, which significantly boosted memory chip prices [5] - Eight government departments issued a document to promote the development of intelligent chips, emphasizing the need for advancements in key technologies such as high-end training chips and AI servers [6] Group 1: ETF Performance - The semiconductor ETF Tianhong (159310) has seen a scale increase of 32.55 million yuan over the past two weeks, indicating significant growth [1] - The Tianhong Science and Technology Innovation Index ETF (589860) reached a new high of 313 million yuan, marking its best performance in the last month [2] - The underlying indices for both ETFs reflect strong market activity, with notable gains in constituent stocks such as Changchuan Technology (300604) and Yinuosi (688710) [1][2] Group 2: Market Trends - The DRAM market is experiencing an upward cycle, with a projected increase of 55-60% in contract prices for general DRAM in Q1 2026, driven by supply constraints due to the shift of production to meet AI server demands [6] - The increase in memory chip prices is expected to significantly improve the profitability of domestic storage leaders, as noted by Shengan Securities [6] - The government's focus on enhancing AI computing power and supporting intelligent chip development is likely to create new investment opportunities in the technology sector [6]