Core Viewpoint - DeepSeek has launched its updated model DeepSeek-V3.2-Exp, which significantly reduces API costs for developers by over 50% due to lower service costs associated with the new model [1][9]. Model Release and Features - The DeepSeek-V3.2-Exp model was officially released on September 29 and is available on the Hugging Face platform, marking an important step towards the next generation architecture [3]. - This version introduces the DeepSeek Sparse Attention (DSA) mechanism, which optimizes training and inference efficiency for long texts while maintaining model output quality [5][8]. - The model supports a maximum context length of 160K, enhancing its capability for handling extensive data [4]. Cost Structure and API Pricing - The new pricing structure for the DeepSeek API includes a cost of 0.2 yuan per million tokens for cache hits and 2 yuan for cache misses, with output priced at 3 yuan per million tokens, reflecting a significant reduction in costs for developers [9]. Open Source and Community Engagement - DeepSeek has made the DeepSeek-V3.2-Exp model fully open source on platforms like Hugging Face and ModelScope, along with related research papers [11]. - The company has retained API access for the previous version, V3.1-Terminus, to allow developers to compare performance, with the same pricing structure maintained until October 15, 2025 [11]. Upcoming Developments - There are indications that the new model GLM-4.6 from Z.ai will be released soon, which is expected to offer greater context capabilities [15][16].

Model Release and Architecture - DeepSeek officially released and open-sourced the DeepSeek-V3.2-Exp model on September 29 [1] - The model introduces a sparse Attention architecture, effectively reducing computing resource consumption and improving model inference efficiency [1] Deployment and Infrastructure - The model is now available on Huawei Cloud's Model as a Service (MaaS) platform [1] - Huawei Cloud continues to use the large EP parallel scheme for DeepSeek-V3.2-Exp deployment [1] - The deployment leverages a context-parallel strategy with long sequence affinity based on the sparse Attention structure, balancing model latency and throughput performance [1] Pricing and Cost Reduction - DeepSeek has significantly reduced the cost of the new model service, leading to a corresponding decrease in official API prices [1] - Developers can expect a cost reduction of over 50% when calling the DeepSeek API under the new pricing policy [1]

21世纪经济报道记者 张晓慧 深圳报道 一是强化品牌赋能，推动黄金珠宝企业与品牌策划机构对接，提高企业品牌影响力，同时吸引国内外品 牌珠宝企业设立全国总部、区域总部、电商总部。 9月29日，深圳召开发布会，介绍深圳推动传统产业向"新"向"智"，加快建设现代化产业体系的有关情 况。发布会对深圳推动传统产业转型升级的12条举措进行解读，并发布了"中国宝都深圳水贝"的下一步 发展规划与布局。 据悉，深圳黄金珠宝行业年工业产值连续多年稳居全国首位，全年黄金提货量约占上海黄金交易所的 70%。 深圳水贝作为全国最大的黄金珠宝集散地之一，培育了周大生、周六福等上市企业；其所在的罗湖区集 聚了近8000个黄金珠宝产业法人企业，2024年，区内黄金珠宝产业实现增加值88.66亿元，增长4.6%。 罗湖区副区长时冬昕介绍，未来罗湖将主要从五个方面入手，引导黄金珠宝产业朝高质量发展，系统推 动罗湖黄金珠宝产业从"规模制造"向"价值创造"的跃升，进一步擦亮"中国宝都、深圳水贝"的金字招 牌。 具体来看： 二是推进数智赋能，据悉，罗湖将联合华为云建设黄金珠宝产业数智化赋能中心，推动AI大模型与珠 宝产业结合，聚焦"AI+设计研发"" ...

Core Viewpoint - Shenzhen is actively promoting the optimization and upgrading of traditional industries, leveraging innovation, digital transformation, and international market expansion to build a modern industrial system [1][2][3][4][5][6][7] Group 1: Achievements in Traditional Industries - Shenzhen's traditional industries have shown significant achievements, with leading brands in high-end women's clothing, jewelry, furniture, and eyewear, contributing to a robust industrial foundation [2] - The jewelry industry maintains the highest industrial output value in China, with annual gold delivery volume accounting for approximately 70% of the Shanghai Gold Exchange [2] - The furniture sector is recognized as a key research, design, production, and export base, while the eyewear industry accounts for about half of global production [2] Group 2: Innovation and Digital Transformation - The upgrade of traditional industries in Shenzhen is closely tied to innovation, particularly through the integration of AI and digital technologies [3] - Shenzhen plans to implement AI across various sectors, including home appliances and toys, establishing a comprehensive technological foundation [3] - Aiming to support over 100 traditional enterprises in digital transformation within three years, Shenzhen is focusing on building industrial internet platforms and solution providers [3] Group 3: Brand and Quality Enhancement - Shenzhen is committed to enhancing product value through brand development and quality improvement, with initiatives to create over 100 cross-industry fashionable and trendy products by 2027 [4] - The city will promote the establishment of advanced group standards in key areas like smart home and eyewear, aiming to elevate industry standards [4] Group 4: International Market Expansion - Shenzhen is focused on enhancing the global presence of "Shenzhen design" and "Shenzhen manufacturing" through various support measures for internationalization [5][6] - The city provides one-stop services for enterprises to expand internationally, including participation in nearly 300 overseas trade fairs annually [6] Group 5: Ecosystem and Support Mechanisms - To ensure the implementation of initiatives, Shenzhen is establishing a traditional industry working group to coordinate development and provide unified services [7] - Local districts are fostering collaboration between industries and educational institutions, enhancing talent retention and cultural product richness [7]

Core Insights - The competition for dominance in China's "AI cloud" market is intensifying as companies like Alibaba Cloud, ByteDance's Volcano Engine, and Baidu Smart Cloud vie for market share [1][5] - The definition and scope of "AI cloud" remain ambiguous, leading to varying statistics and claims from different cloud service providers [2][6] Market Research Reports - Seven market research reports on the AI cloud market have been published by IDC, Omdia, and Sullivan, with IDC providing the most comprehensive analysis [3][4] - Omdia estimates the total revenue of China's AI cloud market to be $3.1 billion (approximately 221 billion yuan) for the first half of 2025, with Alibaba Cloud holding a 35.8% market share [7] Company Rankings - Alibaba Cloud ranks first in five of the seven reports, while Volcano Engine leads in the rapidly growing MaaS model service market [5][14] - Baidu Smart Cloud and Alibaba Cloud are tied for first place in the AI public cloud services market, which has evolved from the small model era prior to 2022 [15] Market Growth and Projections - Alibaba Cloud's revenue for the first half of 2025 is projected to be 63.5 billion yuan, reflecting a year-on-year growth of 21.8% [28] - Volcano Engine aims for revenue exceeding 20 billion yuan in 2025, with a long-term goal of surpassing 100 billion yuan by 2030 [29] Token Market Dynamics - The token calling market is currently small but is expected to grow rapidly, with daily token consumption in China increasing from 100 billion to over 30 trillion in just over a year [30][31] - The revenue from token calls is projected to grow exponentially, making it a critical area for competition among cloud service providers [32][33]

云栖大会上，吴泳铭预测，未来五年全球AI累计投入金额将超过4万亿美元，"这是人类历史上最大的算 力和研发投入"。对比2022年生成式AI元年，2032年阿里云全球数据中心的能耗规模将提升10倍。 吴泳铭说，阿里云是少数能做到软硬件垂直整合的AI云计算平台。在硬件和网络层面，阿里云自研核 心存储系统、网络架构、计算芯片等；阿里云同时拥有业内领先的通义模型，从而实现基础设施架构设 计与模型架构设计的协同创新。 云栖大会上，阿里云智能集团首席技术官周靖人发布了7款通义大模型，以及全新升级的阿里云多款AI Infra产品和服务。 "行业对AI基础设施的需求远超我们的预期。"9月24日的2025云栖大会上，阿里巴巴集团董事兼首席执 行官吴泳铭透露，阿里正在推进3800亿元的AI基础设施建设，并计划追加更大的投入。 受此消息影响，截至9月24日港股收盘，阿里巴巴（09988.HK）股价大涨9.16%，报174.0港元/股，创下 近四年来新高。 此前2月，吴泳铭曾宣布，阿里将在未来三年投入超过3800亿元，用于建设云和AI硬件基础设施，投资 总额超过去十年的总和。 AI基建（AI Infra）之于AI，如同电⽹之于电⼒。云 ...

Core Insights - Canalys has released the "2025 China Cloud Channel Leadership Matrix," recognizing Alibaba Cloud, Huawei Cloud, and AWS as the "Champions" for their outstanding performance in channel management, market performance, and ecosystem development [1][7]. Evaluation Criteria - The matrix evaluates seven major cloud service providers in mainland China with annual cloud service revenues exceeding $1 billion, focusing on vision and strategy, channel planning and execution, partner feedback, and AI ecosystem competitiveness [3]. - The inclusion of AI ecosystem competitiveness in the evaluation highlights the rapid transformation of the channel ecosystem driven by AI, which raises the bar for partner capability building and value delivery [3]. Champion Characteristics - Champions have demonstrated exceptional performance in developing the industry channel ecosystem and have made significant progress in corporate strategy over the past 12 months, leading to recognition from both Canalys analysts and channel partners [3]. - To become a leader in the ecosystem, companies must exhibit three key capabilities: providing precise and continuous partner empowerment, lowering entry barriers, and helping partners accumulate practical experience [4][6]. Market Dynamics - The rapid evolution of AI has propelled the Chinese cloud service market into a new phase, reshaping the role of partners from mere executors of market expansion and resource delivery to becoming critical bridges between vendors and customers [3]. - Leading vendors are actively building AI ecosystems and launching various empowerment initiatives, including targeted incentives, joint solutions, and professional training, to help partners excel in the new competitive landscape [3]. Other Vendor Categories - Vendors are categorized into different groups based on their channel strategy and leadership: - Challengers (Contenders) possess advantages in industry leadership and channel strategy - Expanders (Scalers) show strong development momentum but have limited overall leadership - Foundations (Foundation) lag in partner support and channel planning [5]. Long-term Growth Potential - The long-term growth potential of the ecosystem depends on several key factors, including the ability of vendors to conduct sustained and precise empowerment, partners' willingness to invest in sustainable capability building, and the establishment of stable, trust-based cooperative relationships [6][13].

近日，未可知人工智能研究院院长杜雨博士应邀为 国家能源集团 开展主题为 "AI赋能能源行业智能化转型：创新趋势与实践应用" 的企业内训。此次培 训聚焦人工智能在能源行业的最新发展趋势、核心技术突破及落地应用场景，吸引了集团各业务板块的管理与技术骨干参与。 培训中，杜雨博士结合 国家电网、南方电网、华为云、施耐德电气 等国内外典型案例，深入剖析了AI在智能巡检、功率预测、客服系统、能源管理等场 景中的成熟应用。他特别提到，由未可知人工智能研究院持续关注的国产大模型代表——DeepSeek，正以其高效低成本的训练优势，加速AI在能源等垂 直行业的落地进程。 杜雨博士从宏观、中观、微观三个维度系统讲解了AI如何重塑能源产业格局。他指出，当前全球AI产业正处于快速发展期，生成式AI作为新一轮技术革 命的核心引擎，正深刻改变传统行业的生产方式与服务模式。 尤其在能源领域，AI不仅在资源勘查、发电调度、电网运维等环节实现降本增效，更在推 动"双碳"目标落地、构建新型电力系统中发挥关键作用。 " AI不是未来的技术，而是当下的生产力。 "杜雨博士强调，能源企业应积极拥抱AI技术，从顶层设计到业务场景逐步推进智能化转型，构建 ...

人民财讯9月23日电，贵广网络（600996）9月23日发布股票交易风险提示公告，公司关注到市场有关贵 广网络与省国资体系内有关酒企、云上贵州企业重组，以及公司与华为云开展重大算力合作相关传闻。 经自查，公司不涉及上述情形，市场关于公司的上述传闻不属实。 ...

Group 1 - Shengshu Technology has completed a Series A financing round amounting to hundreds of millions [1] - Dahong Technology has signed agreements with Alibaba Cloud and Yundenshu Technology [1] - Tencent has launched a professional-grade AI 3D workstation called Hongyuan 3D Studio [1] Group 2 - Huawei Cloud CEO Zhang Pingan announced the full launch of CloudMatrix384 AI Token services [1] - Huawei Cloud's embodied intelligence "R2C protocol" has completed the establishment of three national standards [1]