Group 1 - The semiconductor industry is experiencing significant growth, with various indices showing substantial increases, such as the CPO index rising over 123% from April 8 to August 21 [3] - Longjing Technology has completed a strategic financing round, attracting investments from notable funds, indicating strong investor interest in the semiconductor sector [2] - Lante Optics reported a 52.54% increase in revenue and a 110.27% increase in net profit for the first half of 2025, driven by demand in consumer electronics and automotive sectors [3] Group 2 - The release of DeepSeek-V3.1, designed for next-generation domestic chips, highlights advancements in technology within the semiconductor industry [2] - Open-source securities suggest that the uncertainty caused by the H20 embargo has led to a temporary slowdown in capital expenditures for internet companies, but this is expected to recover as project tenders restart [4] - The semiconductor materials ETF and its related funds focus on companies in the semiconductor equipment and materials sectors, which are crucial for domestic substitution and are expected to benefit from the expansion of semiconductor demand driven by AI [4]

2、科创板补涨逻辑：2025陆家嘴论坛上，高层提出在科创板设置科创成长层，重点服务AI等前沿科技 企业。中信证券指出，自今年4月以来明显滞涨的科创板有望迎来补涨行情。 3、国产替代逻辑：招商证券表示，出于对英伟达芯片安全漏洞的担忧，国内部分客户对于未来购买英 伟达芯片还会反复权衡。建议长期持续关注华为昇腾、寒武纪和海光信息等国产算力芯片体系。 4、端侧AI大发展逻辑：人形机器人、智能驾驶等终端落地在即，2025年或成为端侧AI硬件换机元年。 申万宏源类比消费电子发展路径，指出AI终端（如AI手机、具身机器人）换机潮有望复制2010年代智 能手机行情，带动产业链估值中枢上移30-50%。 8月21日，DeepSeek-V3.1正式发布。本次升级主要体现为三方面：①混合推理架构：一个模型同时支持 思考模式与非思考模式；②更高的思考效率：相比此前版本，DeepSeek-V3.1-Think能在更短时间内给 出答案；③更强的Agent能力：在工具使用与智能体任务中的表现，有较大提升。 业内人士指出，此前，国产大模型DeepSeek-R1弯道超车，打破海外高端算力封锁，奠定了国产AI企业 后来居上的基石。此次，Deep ...

Group 1 - The core viewpoint of the news is the launch of DeepSeek-V3.1 by DeepSeek Company, which significantly enhances capabilities in the pharmaceutical retail and commercial sectors through advanced AI models [1] - DeepSeek-V3.1 features a hybrid reasoning architecture that allows dynamic switching between "thinking mode" and "non-thinking mode," improving interaction experience and operational decision-making [1] - The non-thinking mode accelerates intent recognition and understanding, reducing response latency, while the thinking mode generates detailed reports for complex tasks, aiding business operations [1] Group 2 - The 18th West P Conference, a major event in the health industry, was held on August 17 in Hainan Boao, showcasing various intelligent models from Zhongkang Technology [2] - Zhongkang Technology presented five industry-specific intelligent models, including pharmacy and commercial intelligent agents, highlighting its leading position in the healthcare AI sector [2] - The immersive experience zones for pharmacy and commercial intelligent agents attracted significant attention from attendees, demonstrating the company's innovative capabilities in the healthcare AI field [2]

Market Overview - The market experienced fluctuations with mixed performance across the three major indices. The Shanghai Composite Index rose by 0.13%, while the Shenzhen Component Index and the ChiNext Index fell by 0.06% and 0.47%, respectively. The total trading volume in the Shanghai and Shenzhen markets reached 2.42 trillion yuan, an increase of 158 billion yuan compared to the previous trading day, marking the seventh consecutive day of trading volume exceeding 2 trillion yuan [1]. Domestic Computing Power Sector - CITIC Securities highlighted the release of the DeepSeek-V3.1 model, which now supports FP8 precision and domestic chips, potentially accelerating the domestic computing power ecosystem [2]. - Tencent's earnings report indicated a diversification in the supply chain for inference chips, suggesting that domestic inference computing power chips may become a competitive advantage amid international supply chain fluctuations [2]. - Huawei's Ascend chips are gradually expanding their applications in government and financial sectors, further demonstrating the increasing market competitiveness of domestic chips [2]. AI Server and Electronic Fluids - Guojin Securities noted that the demand for electronic fluorinated liquids is rapidly increasing due to the rising shipment volumes of AI servers, particularly with the enhanced computing power of NVIDIA's GB300 NVL72 servers [3]. - Various existing liquid cooling solutions, including cold plate, immersion, and spray cooling, are heavily reliant on fluorinated liquid materials. The trend towards efficient thermal interface materials (TIM) and liquid metals is expected to enhance the advantages of two-phase cold plate cooling solutions [3]. - The liquid cooling solution is anticipated to become the mainstream cooling method for future AI servers, prompting a focus on domestic leading companies in the upstream raw materials for fluorinated refrigerants and electronic fluorinated liquids, as well as related enterprises in efficient TIM materials [3]. Commercial Aerospace Industry - Galaxy Securities projected that the commercial aerospace industry in China is expected to experience significant growth, with key technologies such as reusable rockets and low-cost satellites making breakthroughs [4]. - Predictions indicate that within the next three years, the commercial aerospace sector will reach a peak moment, with transportation costs for rockets dropping below 30,000 yuan per kilogram, approaching the technological capabilities of overseas competitors [4]. - The demand side is bolstered by recent policy measures in Guangdong province, which have included low-altitude economy and space tourism as initial demonstration scenarios, expanding the imagination of application scenarios [4]. - Breakthroughs in low-altitude economy applications have been observed in urban air traffic, industrial applications, and agricultural production, alongside anticipated growth in space tourism demand, suggesting further enhancement of industry space [4].

Industry Insights - The Chinese government will initiate the central frozen pork reserve collection to stabilize the pig market, as the national average pig-to-grain price has dropped below 6:1, entering a warning zone [1] - DeepSeek has launched its latest language model, DeepSeek-V3.1, which incorporates a hybrid reasoning architecture and enhanced agent capabilities, intensifying competition in the global AI market [2] - A new policy financial tool with a scale of 500 billion yuan will be introduced, focusing on emerging industries and infrastructure, including digital economy and artificial intelligence [3] - The US and EU have reached an agreement on a trade framework, with the US imposing tariffs of no more than 15% on most EU goods, while the EU plans to purchase US energy products worth up to $750 billion by 2028 [4] Market Trends - Over 20 titanium dioxide production companies have announced price increases, with expectations for a price rise as the peak season approaches, as current profit margins are unsustainable [9] - The Jiangsu Free Trade Zone has received approval for a comprehensive opening of the biopharmaceutical industry chain, presenting significant opportunities for the innovative drug sector [10] - Zhiyuan Robotics has launched the "Zhiyuan A Plan," aiming to create a trillion-level industrial ecosystem within three years, indicating a potential acceleration in the embodied intelligence industry [11] Company Performance - GoerTek reported a net profit of 1.42 billion yuan for the first half of the year, reflecting a year-on-year increase of 15.65% [14] - SANY Heavy Industry achieved a net profit of 5.216 billion yuan in the first half of 2025, marking a 46% year-on-year growth [14]

每经记者｜杨建 每经编辑｜肖芮冬 （一）重要市场新闻 1、美股三大指数均小幅收跌，标普500指数跌0.4%，道指跌0.34%，纳指跌0.34%；大型科技股多数下 跌，特斯拉、Meta跌超1%，苹果、微软、英伟达、奈飞、亚马逊、AMD、英特尔小幅下跌，谷歌小幅 上涨。中概股多数上涨，纳斯达克中国金龙指数涨1.35%，小鹏汽车涨近12%，蔚来涨超9%，老虎证 券、名创优品涨逾6%，爱奇艺、小牛电动涨逾2%；阿特斯太阳能跌逾18%，禾赛跌11%，哔哩哔哩跌 超6%，晶科能源跌逾4%，大全新能源、阿里巴巴跌超1%。 2、国际贵金属期货收盘涨跌不一，COMEX黄金期货跌0.15%，报3383.5美元/盎司；COMEX白银期货 涨0.87%，报38.1美元/盎司。国际油价集体收涨，美油主力合约涨1.23%，报63.48美元/桶；布伦特原油 主力合约涨1.14%，报67.60美元/桶。欧洲三大股指收盘涨跌不一，德国DAX指数涨0.07%报24293.34 点，法国CAC40指数跌0.44%报7938.29点，英国富时100指数涨0.23%报9309.2点。 （二）行业掘金 拟于2026年在南海启动首个万吨级产能示范基地，预 ...

8月19日晚间，DeepSeek官方悄然上线了全新的V3.1版本。 官方公告强调了上下文长度拓展至128k，但随着社区的深入挖掘和实测，这次"小更新"之下其实有着更多模型架构的变革和模型重点能力的微调，在编程 能力上的提升也可圈可点，成本优势重回显著。 然而，模型融合的技术路线也引发激烈争论，部分用户反馈旧版模型的"顽疾"复现，对这次更新的评价呈现出两极分化的态势。 发布两天后，DeepSeek官方在公众号上发布了相关消息。 此时，正适合我们回看V3.1，更细致地拆解这次"小更新"。 1 架构之变：V3.1吞掉R1，减轻部署复杂度 尽管DeepSeek官方在更新通知中将"上下文长度拓展至128k"作为核心亮点，但此前的V3版本早已支持128K上下文，只是官方API接口此前仅开放至64K。 因此，这次更新的真正核心并非上下文长度，而是模型底层的架构演进。 根据官方最新发布的公众号内容，确认了V3.1为混合推理架构，即使用一个模型同时支持思考模式与非思考模式。 目前在DeepSeek的官方网页和APP上，即使用户开启"深度思考"模式，模型的标识也已从过去的"R1"变为了统一的"V3"。 除了架构改变外，V3.1被 ...

Group 1 - DeepSeek's latest V3.1 version utilizes UE8M0 FP8 Scale parameter precision, designed for the upcoming domestic chip release [1] - FP8 is a cutting-edge low-precision format for AI computing, significantly enhancing GPU performance and reducing memory usage for large language model training [1] - Domestic GPUs are rapidly developing, transitioning from "usable" to "user-friendly" stages, although they have not yet matched international products [1] Group 2 - Companies like Cambrian, Haiguang Information, and Huawei are leading the A-share computing chip market [3] - Moer Thread provides AI training and inference cards, with its latest GPU supporting FP8 precision, significantly boosting AI computing power [1] - Muxi offers C series GPUs for integrated training and inference, and N series GPUs focused on cloud AI inference, showcasing strong mixed-precision computing capabilities [2] Group 3 - The global GPU market is projected to reach 36,119.74 billion yuan by 2029, with China's GPU market expected to reach 13,635.78 billion yuan, increasing its global market share from 30.8% in 2024 to 37.8% in 2029 [2] - DeepSeek is driving the shift of AI applications from centralized cloud services to mass terminals, necessitating high-cost performance dedicated chips [2] - The domestic chip manufacturers and application enterprises are accelerating their integration with DeepSeek, anticipating a significant increase in domestic computing power by 2025 [2] Group 4 - Huawei's Ascend ecosystem includes companies like Tuo Wei Information, Digital China, and Huafeng Technology, enhancing mixed inference architecture and agent capabilities [4] - The upgraded model shows significant improvements in tool usage and intelligent agent tasks through Post-Training optimization [4] Group 5 - Related companies include Dingjie Zhizhi, Fanwei Network, and Kute Intelligent [5]

Group 1 - The "2025 Fortune China Tech 50" list highlights Huawei, DeepSeek, and CATL as the top three companies, recognizing those that are "born in China and influence the world" [1] - Huawei continues to be a dominant player in the tech sector, while DeepSeek's rise is a notable highlight, achieving a score of 88.5 in the MMLU benchmark test with its DeepSeek-R1 model [1][3] - DeepSeek has surpassed 163 million monthly active users, positioning itself as a leader in the global AI-generated content application market [1] Group 2 - Emerging companies like Unitree Technology and Yunsen Technology are gaining recognition alongside traditional tech giants, with Unitree achieving a global sales volume of 18,000 units and a market share of 23% in quadruped robots [3] - Yunsen Technology focuses on humanoid and quadruped robots, with its "Mountain Cat" all-terrain robot being widely applied across various sectors [3] - The list's changes reflect the vigorous development of China's tech industry and indicate an intensifying future competition in the tech sector [5]

Core Insights - DeepSeek officially released DeepSeek-V3.1 on August 21, 2023 [1] - The upgrade includes significant enhancements in search evaluation metrics, particularly in complex search tests and multidisciplinary expert-level challenges [3][4] Upgrade Features - The official app and web model have been upgraded to DeepSeek-V3.1, allowing users to switch between thinking and non-thinking modes via a "Deep Thinking" button [2][3] - DeepSeek API has also been upgraded, with deepseek-chat corresponding to non-thinking mode and deepseek-reasoner corresponding to thinking mode, expanding context to 128K [3] - The API Beta interface supports strict mode Function Calling to ensure outputs meet schema definitions [3] Performance Improvements - DeepSeek-V3.1 shows significant performance improvements over R1-0528 in complex search tests (browsecomp) and multidisciplinary expert-level problem tests (HLE) [3][4] - The model utilizes UE8M0 FP8 Scale parameter precision, with substantial adjustments made to the tokenizer and chat template, resulting in noticeable differences from DeepSeek-V3 [3] Market Reaction - DeepSeek concept stocks experienced a sharp rise in the market towards the end of the trading day [3]