2025年国庆期间，国产AI算力生态持续演进：服务器订单方面，工商银行、中国联通公布合计百亿服 务器招标结果，国产算力供应商中标占比超90%；大模型方面，阿里发布大模型Qwen3-VL-30B-A3B且 华为昇腾实现0 day支持，腾讯混元最新视觉模型跻身LMArena榜单全球第三。 中信证券指出，当前国产大模型正加速迭代，而国产算力芯片实现无缝适配，实现生态闭环，看好其推 动国产AI持续发展。此前，在2025第二季度海外AI芯片由于地缘政治而受限的背景下，阿里等云厂商 Capex仍能高速增长，一方面国产AI芯片持续迭代，自主可控进展顺利，保障了算力基础设施的持续扩 张；一方面国内云厂商已展现出加速追赶北美AI厂商的决心，预计后续将有更多云厂商跟进投入，带 动国内算力重回高速增长轨道。因此，建议重视国产算力的投资机遇。 ...

Core Insights - DeepSeek has officially upgraded its model to DeepSeek-V3.1-Terminus, enhancing performance based on user feedback and improving language consistency and agent capabilities [1][2] - The new model shows improved stability in output, with benchmark results indicating performance increases in various assessments compared to the previous version [1] - The release of DeepSeek V3.1 is seen as a significant breakthrough for domestic large models and chip ecosystems, reducing reliance on NVIDIA standards and promoting domestic computing power autonomy [2][3] Model Performance - The benchmark results for DeepSeek-V3.1-Terminus show improvements in several areas, including: - MMLU-Pro: 84.8 to 85.0 - Humanity's Last Exam: 15.9 to 21.7 - SimpleQA: 93.4 to 96.8 - BrowseComp: 30.0 to 38.5 [1] - The model's agent capabilities have significantly improved, which is expected to enhance commercial applications of AI agents [3] Industry Impact - The launch of DeepSeek V3.1 has led to a surge in the domestic computing industry, with increased demand for AI chips and related infrastructure [3][4] - The success of DeepSeek is viewed as a victory for open-source models, prompting other Chinese companies to adopt similar open-source strategies [3] - The AI computing demand is projected to grow, benefiting various segments of the computing industry, including AI chips, servers, and related technologies [4] Related Companies - Baidu has released its Wenxin model X1.1, showing significant improvements in performance metrics compared to previous versions and competing models [6] - Alibaba's Tongyi Qianwen has launched the Qwen3-Max-Preview model, marking advancements in the domestic large model sector [6] - SenseTime's new interactive platform integrates with Xiaomi AI glasses, showcasing the application of AI in real-world scenarios [7] - ZTE has introduced several products focused on AI and intelligent computing, facilitating the deployment of DeepSeek models across various industries [7]

Investment Rating - The industry investment rating is "Positive" and is maintained [7]. Core Insights - The computer sector rebounded last week, increasing by 3.47%, ranking 6th among major industries in the Yangtze River region, with a trading volume accounting for 7.79% of the total market. The rebound followed a significant previous decline [2][4][16]. - The release of the Qwen3-Next open-source model by Alibaba is expected to significantly reduce costs and accelerate the implementation of AI applications, showcasing advancements in domestic large models [6][42]. - The report suggests focusing on the Chinese inference computing industry chain, particularly recommending the domestic AI chip leader, Cambricon, as well as the Alibaba Cloud ecosystem, cloud service providers, and IDC collaborations with major companies like Tencent and ByteDance [6][42]. Summary by Sections Market Performance - The computer sector experienced a rebound with a 3.47% increase, while the Shanghai Composite Index rose by 1.52%, closing at 3870.60 points [4][16]. - The trading volume of the computer sector represented 7.79% of the total market, indicating active trading in computing-related stocks [2][16]. Key Developments - The Ministry of Transport issued guidelines for the construction of a "Transportation Power" initiative, which is expected to drive investment opportunities in transportation information technology [21][24]. - The Ministry of Commerce initiated an anti-discrimination investigation against the U.S. regarding measures affecting China's semiconductor industry, which may create investment opportunities in domestic AI chips [27][32]. Recommendations - The report emphasizes the importance of the Qwen3-Next model's release, which is anticipated to enhance the performance and reduce training costs of AI applications, thereby boosting demand for computing power [6][42]. - Investors are advised to pay attention to companies with technological reserves in transportation information technology and those involved in low-altitude and vehicle-road-cloud integration [26][27].

Investment Rating - The report maintains a "Positive" investment rating for the industry [7] Core Insights - On September 12, Alibaba released the next-generation foundational model architecture Qwen3-Next and open-sourced the Qwen3-Next-80B-A3B series models, showcasing significant breakthroughs in model architecture that demonstrate the continuous advancement of domestic large models towards world-leading levels [2][4] - The performance improvements of Qwen3-Next come with a substantial reduction in training costs, which is expected to accelerate the deployment of domestic AI applications and drive a surge in computing power demand [2][4] Summary by Sections Event Description - The report details the release of Qwen3-Next and its open-sourced model series on September 12, highlighting the advancements in model architecture [4] Model Innovations - Qwen3-Next features several innovations, including: 1. A mixed attention mechanism that combines Gated DeltaNet and standard attention to balance performance and efficiency 2. A high sparsity MoE structure with a total parameter count of 80 billion, activating only about 3 billion parameters during inference 3. Stability optimizations to prevent weight growth and ensure numerical stability 4. A multi-token prediction mechanism that enhances overall model performance [9] Recommendations - The report suggests focusing on: 1. The Chinese inference computing power industry chain, particularly recommending domestic AI chip leader Cambricon 2. Alibaba Cloud's industry chain 3. Cloud service providers 4. IDC, with a focus on collaborations among major companies like Tencent, Alibaba, and ByteDance [2][9]

Core Viewpoint - The global AI code generation competition is experiencing a significant shift, with OpenAI's GPT-5 series models gaining strength while Anthropic's position is weakening due to internal issues and external competition [1][4]. Group 1: Competitive Landscape - Anthropic's models, including Claude Opus 4.1 and Opus 4, have been acknowledged to have reduced capabilities, leading to a decline in their competitive edge [1]. - OpenAI's GPT-5 Pro is being promoted for its superior coding capabilities, indicating a strong market presence [1]. - Domestic AI model manufacturers are launching new models targeting code generation, such as Kimi-K2-0905 and Qwen3-Max-Preview, which emphasize performance improvements in programming tasks [2][6]. Group 2: Technical Advancements - Kimi-K2-0905 features a context length of 256k and has improved correctness, stability, and logical consistency in long code generation tasks [2][6]. - The model utilizes a Mixture-of-Experts (MoE) architecture with a total of 1 trillion parameters, activating 32 billion during inference, showcasing significant technical capabilities [7][6]. - Kimi-K2-0905 has achieved over 390,000 downloads on Hugging Face in the past 30 days, indicating strong user interest and adoption [3]. Group 3: Pricing Strategy - Kimi-K2-0905 offers competitive pricing for its API, with costs set at ¥1.00 per million tokens for cache hits and ¥4.00 for cache misses, making it an attractive alternative to Anthropic's pricing [17][18]. - The pricing strategy positions Kimi-K2-0905 as a "Chinese alternative" to Claude, maintaining compatibility with Anthropic's API [18][19]. Group 4: Market Integration - Domestic AI manufacturers are increasingly integrating their models into mainstream development tools and applications, enhancing their presence in the market [23]. - The ongoing improvements in performance and user experience are expected to create a positive feedback loop, fostering a more robust application ecosystem and expanding market opportunities [23].

Group 1 - The A-share market experienced fluctuations, with the AI computing sector continuing its pullback, as evidenced by the 5G communication ETF (515050) declining by 0.89% [1] - Despite the overall market trend, certain stocks such as SourceJet Technology, Unisplendour, and others showed resilience, indicating selective investment opportunities [1] - The 5G communication ETF has attracted over 550 million yuan in the last four trading days, bringing its total scale to 9 billion yuan, highlighting strong investor interest in the sector [1] Group 2 - Citic Securities noted that both Alibaba and Nvidia's latest financial reports indicate robust growth in investment from domestic and international CSP manufacturers in computing power [2] - Alibaba's AI-related revenue continues to grow at triple-digit rates, providing a clearer path for AI commercialization and alleviating investor concerns regarding AI investment returns [2] - The recommendation for the computing power sector includes both overseas and domestic computing chains, reflecting confidence in the industry's growth potential [2]

Group 1 - The core viewpoint is that the AI industry is experiencing rapid development, driven by technological revolution and industrial transformation, with significant policy support for AI applications [2] - The AI-focused ETF, Huaxia (589010), has seen a recent decline of 2.73%, indicating a reasonable digestion of previous gains, with a notable liquidity of over 25 million yuan in trading volume [1] - Semiconductor company SMIC reported a revenue of 32.348 billion yuan for the first half of 2025, a year-on-year increase of 23.1%, and a net profit of 2.301 billion yuan, up 39.8%, highlighting the strong performance of the foundry industry [1] Group 2 - The AI industry is expected to accelerate further, with a focus on the development loop from AI infrastructure to large models and applications, supported by continuous policy backing [2] - The Huaxia AI ETF closely tracks the Shanghai Stock Exchange's AI index, covering high-quality companies across the entire industry chain, benefiting from high R&D investment and policy incentives [2] - The foundry industry is characterized by a "stronger becomes stronger" trend, with SMIC being the second-largest pure foundry globally, following TSMC [1]

Group 1 - The A-share market saw a collective rise on August 27, with the Shanghai Composite Index increasing by 0.44%, driven by gains in the communication, electronics, and computer sectors, while beauty care and coal sectors experienced declines [1] - The Chip ETF (159995.SZ) surged by 4.76%, with notable increases in constituent stocks such as Amlogic rising by 12.88%, Lattice Semiconductor by 11.00%, Rockchip by 10.00%, Cambricon by 7.60%, and Zhaoyi Innovation by 6.87% [1] - According to TrendForce, the market share of foreign chips like Nvidia and AMD in China's AI server market is expected to decrease from 63% in 2024 to 42% in 2025, while domestic chip suppliers like Huawei HiSilicon and Cambricon are projected to increase their market share to 40% by 2025 [1] Group 2 - The release of DeepSeek-V3.1 is anticipated to enhance the usability of domestic large models in China, facilitating their application and integration with domestic AI chips, thereby improving the competitiveness of these chips in the market [1] - The Chip ETF tracks the Guozheng Chip Index, comprising 30 leading companies in the A-share chip industry across materials, equipment, design, manufacturing, packaging, and testing, including SMIC, Cambricon, Changdian Technology, and Northern Huachuang [2]

Group 1 - Semiconductor stocks continue their recent upward trend, with notable increases in share prices: SMIC (00981) up 5.52% to HKD 59.3, Jingmen Semiconductor (02878) up 4% to HKD 0.52, Fudan Shanghai (01385) up 2.66% to HKD 34.7, and Huahong Semiconductor (01347) up 2.26% to HKD 54.4 [1] - Deepseek officially released the DeepSeek-V3.1 version, which utilizes UE8M0 FP8 Scale parameter precision, designed for the upcoming generation of domestic chips, indicating potential for large-scale application of domestic chips [1] - Dongfang Securities highlights that the new generation of domestic chips is beginning to support FP8, including models like Muxi 690, Chip Origin VIP9000, Cambricon 690, Moore Thread MUSA 3.1 GPU, and Haiguang Deep Computing No. 3 [1] Group 2 - The firm believes that with advancements in domestic AI chip design technology and manufacturing processes, along with the ongoing development of domestic large models and their compatibility with domestic chips, the market share of domestic computing power is expected to continue increasing [1] - The ongoing bullish trend in domestic chips is anticipated to persist, driven by improvements in technology and market dynamics [1]

Group 1 - Semiconductor stocks continue their recent upward trend, with notable increases in share prices: SMIC up 5.52% to HKD 59.3, GlobalFoundries up 4% to HKD 0.52, Fudan Microelectronics up 2.66% to HKD 34.7, and Hua Hong Semiconductor up 2.26% to HKD 54.4 [1] - Deepseek officially released the DeepSeek-V3.1 version, which utilizes UE8M0 FP8 Scale parameter precision, indicating advancements in the design of next-generation domestic chips that are expected to see large-scale application [1] - Dongfang Securities highlights that the new generation of domestic chips is beginning to support FP8, including models such as Muxi 690, Chip Origin VIP9000, Cambricon 690, Moore Threads MUSA 3.1 GPU, and Haiguang Deep Computing No. 3 [1] Group 2 - The firm believes that with improvements in domestic AI chip design technology and manufacturing processes, along with the ongoing development of domestic large models and their compatibility with domestic chips, the market share of domestic computing power is expected to continue to rise [1] - The ongoing bullish trend in domestic chips is anticipated to persist, driven by advancements in technology and increasing market demand [1]