Investment Rating - The report maintains an optimistic outlook for the AI computing industry in 2026, particularly focusing on Huawei's advancements and opportunities in the sector [3][4]. Core Insights - 2026 is projected to be a pivotal year for the domestic AI computing industry, with Huawei expected to unveil a series of new products and technologies that the market has not yet fully reacted to [3][4]. - Huawei's Ascend 950 series, including the Ascend 950PR and Ascend 950DT chips, is set to significantly enhance AI computing capabilities, with a focus on cost-effective solutions and high performance [4][9]. - The report highlights the importance of storage and connectivity in the evolution of AI computing chips, with Huawei making substantial innovations in these areas [4][9]. - The IPOs of Zhiyuan and MiniMax mark significant milestones in the AI model startup landscape, representing different business models in the sector [3][37]. Summary by Sections Huawei's Computing Innovations - Huawei is positioned as a comprehensive ICT manufacturer, integrating various components from semiconductor supply chains to software ecosystems, which is expected to lead to system-level advancements [4][7]. - The Ascend 950 series is designed to support industry-standard low-precision data formats, significantly improving training efficiency and inference throughput [7][9]. - The upcoming Atlas 950 supernode, based on the Ascend 950DT chip, is expected to achieve unprecedented performance metrics, including an FP8 computing power of 8 EFLOPS and an interconnect bandwidth of 16 PB/s [22][23]. Zhiyuan and MiniMax Comparison - Zhiyuan, backed by Tsinghua University, focuses on B2B localized deployment, achieving rapid revenue growth with a projected revenue of 3.1 billion yuan in 2024, reflecting a year-on-year increase of 150.9% [39][44]. - MiniMax, established in 2021, emphasizes efficient model architecture and rapid commercialization of consumer products, with a significant portion of its revenue coming from overseas markets [50][51]. - Both companies represent distinct paths in the AI model startup ecosystem, with Zhiyuan focusing on enterprise solutions and MiniMax targeting consumer applications [63].

Investment Rating - The report maintains a positive outlook on the computer industry, particularly regarding Huawei's advancements in AI computing capabilities for 2026 [3]. Core Insights - Huawei is expected to play a significant role in the AI computing industry in 2026, with a series of new products and technologies anticipated to be launched, which the market has not yet fully reacted to [3][4]. - The Ascend 950 series, including the Ascend 950PR and Ascend 950DT chips, is set to be released in Q1 2026, featuring significant improvements in performance and cost efficiency [3][9]. - The report highlights the IPOs of Zhiyuan and MiniMax, marking them as representative companies in the large model startup sector, with Zhiyuan focusing on B2B and MiniMax on B2C models [3][38]. Summary by Sections Huawei's Computing Innovations - Huawei's Ascend chip series is planned to include four products over three years, with the Ascend 950 series being the latest [4][5]. - The Ascend 950PR chip will utilize Huawei's self-developed low-cost HBM, significantly reducing investment costs in AI inference and recommendation scenarios [9][23]. - The report emphasizes the importance of storage and computing updates, with Huawei's innovations in AI computing expected to drive system-level advancements [4][7]. Zhiyuan and MiniMax Comparison - Zhiyuan, founded in 2019, focuses on a comprehensive MaaS platform with a strong emphasis on localized deployment for B2B clients, achieving rapid revenue growth [39][40]. - MiniMax, established in 2021, prioritizes efficient model architecture and rapid commercialization of products, with a significant portion of its revenue coming from overseas markets [52][53]. - Both companies are positioned to represent different paths in the large model sector, with Zhiyuan's focus on B2B and MiniMax's emphasis on B2C applications [66]. Key Company Updates - Inspur Information is noted for its strong market share in AI servers, with new product launches aimed at supporting large model training and inference [68]. - TaxFriend's AI digital accounting solutions have exceeded expectations, indicating a strong product-market fit in the evolving AI landscape [70].

Core Viewpoint - China Mobile has awarded a procurement project for a super node testing device to Huawei, with a bid of 135 million yuan, indicating a significant investment in advanced computing infrastructure [1] Group 1: Huawei's Innovations - Huawei's "Scale-up super large-scale super node computing platform" was recognized as its top invention, showcasing a new architecture that allows for flexible resource allocation among AI processors [1] - The Ascend 384 super node features 384 Ascend NPUs and 192 Kunpeng CPUs, achieving a total computing power of 300 Pflops, which is 1.7 times that of NVIDIA's NVL72 [2] - The total network bandwidth of the Ascend 384 super node reaches 269 TB/s, surpassing NVIDIA's NVL72 by 107% [2] Group 2: Industry Trends - The demand for computing power in large models is rapidly increasing, while traditional computing architectures face challenges such as low resource utilization and frequent failures [4] - Super nodes are emerging as a new norm in AI infrastructure, with several domestic manufacturers exploring super node systems [4] - Inspiring innovations include the launch of the world's first single-cabinet 640-card super node by Sugon, which enhances computing density by 20 times compared to traditional solutions [4][5] Group 3: Future Developments - Huawei plans to launch the Atlas 950 super node in Q4 2026, featuring 8192 Ascend 950DT chips and achieving FP8 computing power of 8 EFLOPS [12] - The Atlas 950 super node is projected to be significantly more powerful than NVIDIA's upcoming NVL144, with a total computing power 6.7 times greater and memory capacity 15 times larger [14] - Future plans include the Atlas 950 SuperCluster, which will consist of 64 Atlas 950 super nodes, achieving a total FP8 computing power of 524 EFLOPS, surpassing the current largest cluster, xAI Colossus [14]

Core Viewpoint - Huawei emphasizes the importance of open innovation and intellectual property protection to drive technological advancement and societal development, showcasing its significant R&D investments and achievements in the field of patents and technology [1][3][9]. R&D Investment - Huawei has consistently invested over 20% of its annual sales revenue into R&D, with expenditures exceeding RMB 170 billion in 2024, representing 20.8% of total sales [1]. - Over the past decade, Huawei's cumulative R&D investment has surpassed RMB 1.2 trillion, ranking sixth globally in R&D spending according to the 2024 EU Industrial R&D Investment Scoreboard [1]. Intellectual Property Achievements - In 2024, Huawei publicly disclosed a record 37,000 new patents, contributing over 10,000 technical proposals to domestic and international standards organizations, and publishing more than 1,000 academic papers [5]. - As of the end of 2024, Huawei holds over 150,000 valid patents globally, with more than 50,000 in China, approximately 29,000 in the U.S., and around 19,000 in Europe [5]. Patent Licensing Revenue - Huawei's patent licensing revenue reached a historical high of $630 million in 2024, marking an increase from $560 million in 2022 [6]. - Cumulatively, Huawei's patent licensing revenue over the past decade is estimated to exceed $4 billion, despite having paid approximately $12 billion in licensing fees to others [6][7]. Open Source and Collaboration - Huawei promotes open-source initiatives, with over 8,100 contributors in the OpenHarmony community submitting more than 10 million lines of new open-source code in the past year [3]. - The company has launched various open-source and hardware projects, including the CANN computing framework and the Cangjie programming language, aiming to provide developers with comprehensive tools and support [3]. Industry Collaboration - Huawei has signed 233 patent licensing agreements, with 95% achieved through negotiation and mediation, collaborating with major international firms like Nokia, Ericsson, Amazon, and Samsung [9]. - The company advocates for a balanced intellectual property strategy, emphasizing the importance of both incentivizing innovation and facilitating application [7][9]. Public Access to Patents - Over 200,000 Huawei patents have entered the public domain for free use, and the company has launched a patent query tool to assist global researchers [9][10]. - The "Chasi Patent" project has received over 2 million visits since its launch, reflecting its positive reception among the research community [10].

Group 1 - The industry allocation strategy for October emphasizes high-risk preferences, suggesting that technology growth and non-bank financial sectors are likely to benefit significantly from the implementation of AI-related policies, particularly in downstream applications like media and computing, as well as strong beta stocks in brokerage firms [1] - The expectation of a Federal Reserve interest rate cut is leading to increased attention on non-ferrous metals and basic chemicals [1] - The top five sectors with net inflows include non-ferrous metals, domestic software, rare earth materials, military industry, and wind power [1] Group 2 - Huawei has announced a series of upcoming products aimed at enhancing AI computing power, including the Ascend 950PR/950DT, with launches scheduled from 2026 to 2028, indicating a strong focus on meeting the growing demand for AI computing capabilities [3] - Major overseas storage manufacturers have adjusted prices, with SanDisk increasing flash memory prices by over 10%, and Micron pausing quotes, suggesting a potential rise in NAND prices due to increased enterprise SSD demand [5] - The price of iodine is expected to rise significantly due to emerging demands from perovskite solar materials and solid-state batteries, with companies that have quality iodine recovery capacity and overseas iodine mining rights being highlighted as key investment opportunities [5] Group 3 - The Shanghai Composite Index has reached a new high of 3900 points, indicating a shift from a sideways market in September to a bullish trend in October, with significant inflows of capital driving the market upward [9] - The market is experiencing a rotation from small-cap to large-cap stocks, with the ChiNext Index continuing to rise, suggesting a strong bullish sentiment [9] - Precious and industrial metals have performed well under the backdrop of a weakening dollar and expectations of interest rate cuts, while the bond market remains relatively weak in comparison to the stock market [9]

Investment Rating - The report maintains a "Positive" investment rating for the communication industry [9]. Core Insights - The communication sector has seen a 0.36% increase in the 38th week of 2025, ranking 13th among major industries in the Yangtze River region. Since the beginning of 2025, the sector has surged by 63.61%, leading all industries in the same region [2][4]. - Huawei has announced the Ascend series chip roadmap, introducing the Atlas 950 super node and the Lingqu 2.0 interconnection protocol, aiming to build a cluster of over 520,000 cards, with the Atlas 960 targeting a million-card scale [5][7]. - Microsoft is accelerating the expansion of the Fairwater cluster, investing an additional $4 billion in Wisconsin, bringing the total investment to $7.3 billion, to support AI model training [6][7]. Summary by Sections Market Performance - In the 38th week of 2025, the communication sector rose by 0.36%, ranking 13th among major industries. Year-to-date, the sector has increased by 63.61%, ranking first [2][4]. - Notable stock performances include Dekeli (+59.1%), Changfei Fiber (+35.1%), and Guangku Technology (+27.6%) for gains, while Sanwei Communication (-14.0%), Yongding Co. (-10.8%), and Dongxin Peace (-10.2%) faced declines [4]. Technological Developments - Huawei's Atlas 950 super node, set to launch in Q4 2026, will utilize 8192 Ascend 950 DT chips, achieving a bandwidth of 16.3PB/s, significantly outperforming previous models [5]. - The Atlas 960 super node, expected in Q4 2027, will support up to 15,488 Ascend 960 chips, targeting high-density scenarios with a total computing power of 2 ZFLOPS [5]. Investment Recommendations - The report recommends several companies across various segments: - Telecom Operators: China Mobile, China Telecom, China Unicom - Optical Modules: Zhongji Xuchuang, Xinyi Sheng, Tianfu Communication - Liquid Cooling: Yingweike - Hollow Core Fiber: Fenghuo Communication, Hengtong Optic-Electric - Domestic Computing: Runze Technology, Guanghuan New Network - AI Applications: Boshi Jie, Heertai - Satellite Applications: Huace Navigation, Haige Communication [7].

Core Insights - Huawei's Vice Chairman Xu Zhijun presented a clear development roadmap for its Ascend AI chips, targeting a million-card cluster by 2028, indicating a significant leap in computing power [1][2] - The introduction of the "Lingqu" all-optical interconnection protocol is expected to revolutionize the field of computing interconnectivity, enhancing data transfer speeds and overall efficiency [2] Group 1: Huawei's AI Chip Development - Huawei plans to rapidly iterate its Ascend AI chips, with the Ascend 950 series launching in 2026, followed by the Ascend 960 in 2027, and the Ascend 970 in 2028, aiming for a tenfold increase in overall computing power [1][2] - The Ascend 960 chip will achieve a comprehensive doubling of key metrics such as computing power and memory bandwidth, while the Ascend 970 will further enhance interconnect bandwidth and memory bandwidth [1] Group 2: Super Node Technology - The Atlas 950 super node, built on the new "Lingqu" protocol, is anticipated to maintain a leading position in global computing power for the coming years [2] - The establishment of large-scale computing clusters, potentially reaching 500,000 to 1,000,000 cards, will significantly reduce data transfer delays and enhance computational efficiency [2] Group 3: Implications for AI Development - With robust computing infrastructure, China's AI models are expected to experience explosive growth, positioning them among the global leaders in AI technology innovation and application [2]

Group 1 - A-shares indices opened higher on September 22, with strong performance in the storage chip sector, particularly the Computer ETF (159998) which rose by 2.17% with a trading volume exceeding 47 million yuan [1] - Key stocks in the Computer ETF included Jiangbolong, which increased by over 11%, and other notable gainers were Zhongke Shuguang, Xiechuang Data, and 360 [1] - The Cloud Computing ETF (517390) experienced a slight decline of 0.17%, but notable stocks like Donghua Software reached the daily limit up, with other gainers including Guanghuan New Network and Aofei Data [1] Group 2 - The Cloud Computing ETF (517390) closely tracks the CSI Hong Kong-Shanghai Cloud Computing Industry Index and includes major stocks like Tencent Holdings and Zhongke Shuguang [2] - Recent developments include the establishment of the Shanghai Quantum Artificial Intelligence Joint Alliance and the release of ten application scenarios for quantum computing at the Pujiang Innovation Forum [2] - Huawei announced the "Tian Gong Plan" at the Huawei Connect 2025 conference, committing 1 billion yuan to support the Harmony AI ecosystem, focusing on AI agents as a key direction for development [2] Group 3 - CITIC Securities indicated that the third quarter is a traditional peak season for the electronics sector, with strong demand for AI computing-related products, including chips and servers, and an upward trend in both overseas and domestic computing chains [3] - The demand for storage is expected to increase, with enterprise-level demand remaining robust, and NAND and DRAM demand growth and price increases exceeding expectations since the second half of 2025 [3] - The semiconductor equipment chain is steadily advancing in domestic production, with a clear trend of mergers and acquisitions in the industry [3] Group 4 - CITIC Jinpu highlighted Huawei's upcoming product launches, including the Ascend 950 series, which will be released between 2026 and 2028, aimed at meeting the growing demand for AI computing power [4] - Huawei's Atlas 950 super node, supporting 8192 Ascend 950DT cards, is set to launch in the fourth quarter of 2026, significantly enhancing overall computing power [4] - The North American AI computing demand remains strong, with expectations of a rebound in domestic AI computing demand from September to October, and a recommendation to focus on domestic cloud service providers [4]

Report Industry Investment Rating - The report recommends a long position for IF, IC, IM, and IH in the stock index futures of the macro and financial sector [1]. - Goldman Sachs maintains an overweight rating for A - shares and H - shares [1]. - CITIC Construction Investment Research Report is bullish on the overall Hong Kong stock market [1]. Report's Core View - The Chinese stock market's strong performance this year may be driven by "re - inflation" expectations and artificial intelligence, and future improvements in valuation and liquidity may contribute to further prosperity [1]. - International capital is actively increasing its positions in China's technology sector as China has global competitiveness in frontier fields such as AI, robotics, and biotechnology, and the window for stock market valuation repair is opening [1]. - The narrowing of the Sino - US interest rate spread will attract more global funds to focus on RMB assets, and investors will adjust their asset allocation to increase holdings of Chinese bonds and stocks [1]. - After entering September, the A - share market is in a consolidation period, and the advantages of Hong Kong stocks are emerging [1]. - The current stock market is in a rest stage after a sharp decline on Thursday. The traditional industries are strengthening, and the CSI 300 index stabilizes the market. The market is undergoing a phased style shift from growth to defense, but the upward trend remains unchanged [2]. Summary by Relevant Catalogs Market Review - On Friday, the main indexes of the two markets fluctuated and closed slightly lower, with the CSI 300 index rising. The trading volume of the two markets was 2.32 trillion yuan, showing a rapid contraction. The CSI 300 index closed at 4501 points, up 3 points or 0.08%; the SSE 50 index closed at 2909 points, down 3 points or - 0.11%; the CSI 500 index closed at 7170 points, down 29 points or - 0.41%; the CSI 1000 index closed at 7438 points, down 38 points or - 0.51% [1]. - Among industry and theme ETFs, coal ETF, military industry leading ETF, engineering machinery ETF, real estate ETF, and tourism ETF led the gains, while auto parts ETF, robot 50ETF, and integrated circuit ETF led the losses. Among the sector indexes of the two markets, energy metals, engineering machinery, tourism, film and television theaters, and coal mining indexes led the gains, while motor manufacturing, home appliance parts, auto services, reducers, and PEEK material indexes led the losses [1]. - The settlement funds of stock index futures for the CSI 1000, CSI 500, CSI 300, and SSE 50 indexes had net outflows of 5.8 billion, 3.1 billion, 2.6 billion, and 2.2 billion yuan respectively [1]. Important Information - Goldman Sachs believes that "re - inflation" expectations and artificial intelligence may be the key drivers of the Chinese stock market's strong performance this year, and future improvements in valuation and liquidity may contribute to further prosperity [1]. - Goldman Sachs maintains an overweight rating for A - shares and H - shares, and international capital is actively increasing its positions in China's technology sector [1]. - The narrowing of the Sino - US interest rate spread will attract more global funds to focus on RMB assets [1]. - CITIC Construction Investment Research Report shows that after entering September, the A - share market is in a consolidation period, and the attention of domestic and foreign funds to Hong Kong stocks is increasing, and the advantages of Hong Kong stocks are emerging [1]. - Huawei's core strategy is "super - node + cluster", and its Atlas 950 super - node is leading compared with NVIDIA's planned NVL576 in 2027 [1]. - In July, Japan and the UK increased their holdings of US Treasury bonds, while China reduced its holdings by 25.7 billion US dollars to 730.7 billion US dollars, the lowest level since 2009 [2]. - The Bank of Japan announced that it will sell its ETF holdings at a rate of about 330 billion yen per year and real - estate REIT at a rate of about 5 billion yen per year [2]. - The US Department of Energy launched the "Power Acceleration" plan to meet the surging power demand from artificial intelligence and data centers [2]. - Microsoft will invest 3.3 billion US dollars in a data center in Wisconsin, which will be put into use early next year and will be 10 times more powerful than the current fastest supercomputer [2]. Market Logic - The Chinese stock market's current rally is mainly driven by liquidity, with "re - inflation" expectations and AI autonomy as key catalysts [2]. - If the proportion of foreign institutional holdings in the A - share market rises to the average level of emerging or developed markets, it may bring 14 trillion to 30 trillion yuan of potential capital inflows [2]. Future Market Outlook - The narrowing of the Sino - US interest rate spread will attract more global funds to focus on RMB assets, and investors will increase their holdings of Chinese assets [2]. - The current stock market is in a rest stage after a sharp decline on Thursday. The traditional industries are strengthening, and the CSI 300 index stabilizes the market. The market is undergoing a phased style shift from growth to defense, but the upward trend remains unchanged [2]. Trading Strategy - For stock index futures directional trading, in the rest stage, traditional industries are strengthening, and the CSI 300 index stabilizes the market. The market is undergoing a phased style shift from growth to defense, and the upward trend remains unchanged [2]. - For stock index option trading, during the volatile period when the market is undergoing a phased style shift from growth to defense, it is advisable to wait and see [2].

Core Viewpoint - Huawei is leading a new paradigm in AI infrastructure with its innovative supernode interconnection technology, emphasizing the importance of computing power in artificial intelligence development [2][3]. Group 1: Ascend Chip Development - Huawei has made significant progress in its Ascend chip series, with the upcoming Ascend 950 series expected to enhance AI computing capabilities [3][4]. - The Ascend 950 series includes two chips: Ascend 950PR and Ascend 950DT, which will improve training efficiency and inference throughput [5][6]. - The Ascend 950PR chip targets inference prefill and recommendation scenarios, while the Ascend 950DT focuses on inference decode and training, with significant improvements in memory bandwidth and capacity [6][7]. Group 2: Supernode Products - The Atlas 950 supernode, based on the Ascend 950DT chip, will support 8192 Ascend 950DT chips, significantly increasing computing power and interconnection bandwidth [9][10]. - The Atlas 960 supernode, set to launch in 2027, will further enhance performance with a maximum support of 15488 cards, doubling the capabilities of the Atlas 950 [11][12]. - The Atlas 950 SuperCluster will consist of 64 Atlas 950 supernodes, achieving a total computing power of 524 EFLOPS, making it the world's strongest computing cluster [21][22]. Group 3: Interconnection Technology - Huawei has developed a new interconnection protocol, "Lingqu," to support large-scale supernode architectures, enhancing reliability and bandwidth [20][23]. - The interconnection technology addresses challenges in long-distance, high-reliability connections and aims to achieve TB-level bandwidth with low latency [19][20]. - The Lingqu protocol will be open for industry partners to foster collaboration and innovation in interconnection technology [20][23].