随着壁仞的上市和昆仑芯的冲刺，国产AI算力板块正在资本市场形成新的集聚效应。受市场乐观情绪提振，离岸人民币兑美元汇率继续走高，盘 中一度升破6.97关口、创下2023年5月以来的盘中新高。 截至午盘恒生指数涨幅达2.18%，市场半日成交额为768.69亿港元。"港股GPU第一股"壁仞科技盘中一度涨超118%，截至午盘涨幅收窄至73%。百 度股价大涨超7%，公司宣布分拆旗下AI芯片公司昆仑芯赴港上市。离岸人民币一度升破6.97关口，创下2023年5月以来的盘中新高。 2026年首个交易日，港股市场在科技股带动下强势开局，恒生科技指数涨超2.5%。"港股GPU第一股"壁仞科技上市首日股价飙升，百度因宣布分 拆旗下AI芯片公司昆仑芯赴港上市，股价大涨。 恒生指数高开后持续走高，截至午盘涨幅达2.18%，市场半日成交额为768.69亿港元。恒生科技指数则拉升涨3.38%。港股芯片股集体走强，华虹 半导体涨9.89%，中芯国际涨约5%。 | 序号 代码 | 名称 | 现价 | 涨跌 | 涨跌幅 | 成交额 | | --- | --- | --- | --- | --- | --- | | 11 HSI | 恒生指数 | ...

Group 1 - The report indicates that the market is likely entering a "spring rally" phase, with the overall market represented by the Wind All A Index breaking through moving averages and expected to reach new highs. This is supported by increased trading volume despite the suspension of northbound capital transactions [4][7][13] - The technology growth sector is identified as a long-term market focus, with a K-shaped economic recovery in China favoring technology growth performance. Opportunities are anticipated in previously lagging areas such as AI applications and humanoid robots [4][7][12] - The report highlights the ongoing expansion in the non-ferrous metals market, with prices for gold, silver, and copper reaching historical highs. This is attributed to a combination of factors including a loosening of dollar liquidity and supply-demand tightness [12][13] Group 2 - The report suggests that the domestic commercial aerospace industry is expected to develop rapidly, supported by new regulations that favor commercial rocket enterprises and the anticipated growth of the market, which is projected to exceed 2.5 trillion yuan [11][12] - The report emphasizes the importance of monitoring the non-ferrous metals sector for investment opportunities, particularly in strategic minor metals and industrial metals, as the market is currently in an expansion phase [12][13] - The report notes that monetary policy remains moderately accommodative, which supports a slow bull market in A-shares, with expectations of continued liquidity support for technology growth sectors [8][9]

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].

行 业 及 产 业 2025 年 12 月 27 日 华 为 2026 会 反 转 吗 ？ 智 谱 和 minimax 对比研究！ 看好 ——计算机行业周报 20251222-20251226 相关研究 《字节豆包 1.8 重磅发布！壁仞提交港股 招股书！——计算机行业周报 20251215- 20251219》 2025/12/20 《 OpenAI GPT 5.2 发布！重磅中航信深 度发布！——计算机行业周报 20251208- 20251212》 2025/12/13 证券分析师 黄忠煌 A0230519110001 huangzh@swsresearch.com 洪依真 A0230519060003 hongyz@swsresearch.com 刘洋 A0230513050006 liuyang2@swsresearch.com 研究支持 崔航 A0230524080005 cuihang@swsresearch.com 徐平平 A0230525080002 xupp@swsresearch.com 曹峥 A0230525040002 caozheng@swsresearch.com 陈晴华 A02 ...

其中，江原D20 AI加速卡是国内首款实现设计、制造、封装、境内流片量产的云端AI推理卡，供应链 自主可控，搭载256GB显存，INT8算力峰值达320TOPS，支持FP32/FP16/BF16/INT8多精度混合计算， 为大模型运行提供坚实硬件支撑。 单机性能层面实现关键突破：单卡机型可稳定流畅运行32B参数级大模型，四卡机型凭借多芯片协同架 构与品高自研4D并行调度策略，可单机承载Qwen 235B、DeepSeek 685B等千亿级大模型运行，打破桌 面级工作站超大模型部署的算力限制。 配合一体式液冷散热架构，AI Station系列实现"高性能+低噪音+长稳定"的三重优势：热传导效率较传 统风冷提升40%，可避免高负载下的算力降频，保障7×24小时连续运行；运行噪音低至28dB（接近图 书馆环境音），适配办公、实验室等安静场景。 通过液冷系统的小型化集成，四卡机型仍可保持紧凑机身设计，兼顾算力密度与空间适配性。场景适配 方面，全系列机型深度支持RAG（检索增强生成）、工作流自动化、智能体部署等前沿AI应用场景， 依托本地算力部署模式构建数据私域安全屏障，有效规避核心数据外泄风险，契合政企客户数据合规需 ...

科创人工智能ETF华夏(589010)紧密跟踪上证科创板人工智能指数，覆盖全产业链优质企业，兼具 高研发投入与政策红利支持，20%涨跌幅与中小盘弹性助力捕捉AI产业"奇点时刻"。 每日经济新闻 （责任编辑：张晓波 ） 截至10:10，科创人工智能ETF(589010) 下跌0.96%，盘中震荡调整，回踩支撑位，相对便宜的配置 窗口或已悄然打开。持仓股方面呈现"跌多涨少"的格局。在统计的30只主要持仓股中，仅5只上涨，25 只下跌。澜起科技表现强势，逆势上涨2.63%，复旦微电上涨0.87%，威胜信息、中科星图微涨，显示 出芯片设计端具备一定韧性。软件服务类个股回调明显，星环科技下跌3.62%，合合信息下跌2.95%， 晶晨股份、芯原股份、凌云光等跌幅居前。流动性方面，开盘一小时成交额已逼近3000万元，交投情绪 保持平稳。值得注意的是，该ETF近期"吸金"能力惊人，12月初曾连续多日获资金大幅加仓，合计净流 入超3亿元，显示出"聪明钱"在震荡市中逢低布局的坚定信心，投资者或可关注今日回调带来的低位介 入良机。 华西证券表示，国家政策高度支持，国产AI算力正在崛起根据工信部印发的算力发展规划，2025 年国内智 ...

Group 1 - The core point of the news is the successful listing of domestic GPU company Moore Threads, which saw its stock price surge over 400% and market capitalization exceed 270 billion yuan, marking a significant moment for China's AI chip industry [1] - The rise of domestic GPUs requires not only technological breakthroughs but also the collaboration of software ecosystems, as exemplified by Nvidia's established CUDA software ecosystem [1] - Moore Threads' long-term partnership with SenseTime serves as a model for achieving "soft and hard integration" in domestic GPU development through deep ecological collaboration [1] Group 2 - Moore Threads and SenseTime have maintained a business cooperation for years, with SenseTime's algorithms adapted to Moore Threads' products, enhancing market adaptability [2] - The collaboration focuses on optimizing large model training and inference technologies, with Moore Threads providing full-function GPUs that meet diverse computational needs for SenseTime's applications [2] - This partnership creates a closed-loop cooperation mechanism of "demand-supply-optimization," enhancing market competitiveness for both companies [2] Group 3 - SenseTime's advanced large model system and its AI infrastructure, SenseCore, provide an excellent platform for validating Moore Threads' products, aiding in industry customer expansion [3] - Moore Threads' MTT S series GPUs have been fully adapted to SenseTime's large model system, marking a significant achievement in industrial-grade testing for domestic GPUs [3] - The collaboration allows for performance validation and product refinement, creating a virtuous cycle of ecosystem expansion, customer growth, and performance improvement for both companies [3] Group 4 - In the context of global AI computing supply shortages, establishing a diverse and stable domestic computing supply chain and ecosystem collaboration is crucial [4] - The "SenseTime Computing Mall" connects multiple chip companies to create a domestic AI computing ecosystem, providing comprehensive computing solutions and lowering barriers for enterprises [4] - Moore Threads joined the "SenseTime Computing Mall" as a core member, enabling downstream companies to access validated computing power, thus accelerating market expansion through SenseTime's ecosystem influence [4]

Investment Rating - The report maintains an "Overweight" rating for the computer industry, indicating an expectation that the industry index will outperform the market index by more than 10% over the next six months [1][32]. Core Insights - Domestic technology giants are actively investing in supernode technology, presenting investment opportunities in domestic AI computing power [1][28]. - The SW computer sector has seen a cumulative decline of 2.64% over the past two weeks, underperforming the CSI 300 index by 2.48 percentage points, ranking 28th among 31 first-level industries [10][13]. - The sector's price-to-earnings (PE) ratio is currently at 55.36 times, placing it in the 88.40th percentile for the past five years and the 77.89th percentile for the past ten years [20][22]. Summary by Sections 1. Market Review - The SW computer sector has experienced a cumulative decline of 2.64% from October 31 to November 13, 2025, and a decline of 3.73% in November, while it has increased by 20.45% year-to-date [10][13][28]. 2. Valuation Situation - As of November 13, 2025, the SW computer sector's PE TTM (excluding negative values) stands at 55.36 times, indicating a high valuation relative to historical performance [20][22]. 3. Industry News - Key developments include the launch of the Kimi K2 Thinking model by Moonlight, the introduction of the world's first single-cabinet 640-card supernode by Inspur, and significant investments in AI infrastructure by companies like Tencent and Anthropic [21][23][28]. 4. Company Announcements - Recent announcements include equity transfer agreements and the establishment of joint ventures aimed at enhancing business strategies and market positions [24][25][26][27]. 5. Weekly Perspective - The report highlights the launch of the scaleX640 supernode by Inspur, which significantly enhances computing power and efficiency, and emphasizes the importance of collaborative efforts in the AI computing ecosystem [28][29].

Investment Rating - The report maintains an "Overweight" rating for the computer industry, expecting the industry index to outperform the market index by more than 10% in the next six months [1][32]. Core Insights - Domestic technology giants are actively investing in supernode technology, presenting investment opportunities in domestic AI computing power [1][28]. - The SW computer sector has seen a cumulative decline of 2.64% over the past two weeks, underperforming the CSI 300 index by 2.48 percentage points, ranking 28th among 31 first-level industries [10][13]. - The sector's price-to-earnings (PE) ratio is 55.36 times, placing it in the 88.40th percentile over the past five years and the 77.89th percentile over the past ten years [20][22]. Summary by Sections 1. Market Review - The SW computer sector has experienced a cumulative decline of 2.64% from October 31 to November 13, 2025, and a decline of 3.73% in November, while it has increased by 20.45% year-to-date, outperforming the CSI 300 index by 0.95 percentage points [10][13][14]. 2. Valuation Situation - As of November 13, 2025, the SW computer sector's PE TTM (excluding negative values) stands at 55.36 times, which is in the 88.40th percentile for the last five years and the 77.89th percentile for the last ten years [20][22]. 3. Industry News - Major developments include the launch of the Kimi K2 Thinking model by Moon's Dark Side, the introduction of the world's first single-cabinet 640-card supernode scaleX640 by Inspur, and Tencent's announcement of lower-than-expected capital expenditures for the year [21][23][28]. - Anthropic plans to invest $50 billion in a data center in the U.S., while Baidu has released new Kunlun chips M100 and M300 for AI applications [21][23]. 4. Company Announcements - Recent announcements include a share transfer agreement by Jin Chengzi and a joint venture by Jin Yi Technology to establish a company focused on energy technology [24][25][27]. 5. Weekly Perspective - The report highlights the significance of the scaleX640 supernode, which enhances computing power and efficiency, and emphasizes the importance of domestic AI computing power investment opportunities [28][29].

Core Viewpoint - The semiconductor sector is experiencing a significant boost due to breakthroughs in photoresist technology, leading to strong stock performance among related companies and ETFs [1][4]. Group 1: Stock Performance - Jingrui Electric Materials opened strongly and hit the upper limit, while Nanda Optoelectronics and Aisen Co. saw gains exceeding 12% [1]. - The semiconductor equipment ETF (561980) rose by 2.47%, with a peak increase of over 3%, and recorded a trading volume of 110 million yuan [1]. - Over the past ten trading days, the ETF has seen a cumulative net inflow of 294 million yuan, bringing its latest scale to 2.626 billion yuan, with a year-to-date index increase of 61.31% [1]. Group 2: Company Developments - The research team led by Professor Peng Hailin from Peking University has made advancements in understanding the micro-3D structure of photoresist molecules, which could significantly reduce lithography defects [4]. - Domestic GPU manufacturer Muxi Co. has received approval for its IPO on the Sci-Tech Innovation Board, indicating a positive trend for domestic computing power companies [4]. - Muxi is negotiating customer orders estimated at 2.004 billion yuan, with over 1.7 billion yuan in AI computing projects, while Muxi has an order backlog of 1.43 billion yuan [5]. Group 3: ETF Performance and Holdings - The semiconductor equipment ETF (561980) reported a net asset growth rate of 89.26% over the past year, with a 61.46% increase over the last six months [7]. - The ETF tracks the CSI Semiconductor Index, with top holdings including Zhongwei Company (15.01%), Beifang Huachuang (12.70%), and Cambrian (10.56%), collectively accounting for over 78% of the portfolio [9][10]. - The index is heavily weighted towards semiconductor equipment, materials, and digital chip design, which are expected to benefit from the push for self-sufficiency in the semiconductor industry [9]. Group 4: Financial Performance of Component Stocks - Companies like Haiguang Information and Cambrian reported significant year-on-year growth in both revenue and net profit for the first three quarters [11]. - Jingrui Electric Materials achieved a staggering 192-fold increase in net profit year-on-year, highlighting the strong performance of semiconductor-related firms [12]. - Other component stocks such as Jianghua Micro and Aisen Co. also reported positive revenue and net profit growth, indicating a robust sector performance [12]. Group 5: Market Outlook - Analysts from Huaxi Securities and Galaxy Securities express optimism about the continued growth of the technology sector, driven by advancements in AI and high-performance computing chip demand [6][13]. - The stability of key chip and material supplies is critical, making the establishment of a self-sufficient semiconductor supply chain a necessity rather than an option [13].