2026 年 03 月 04 日 电子 国产算力崛起：内外双轮驱动下的自主 生态突围 内外需求共振，开启国产算力历史窗口 行业深度分析 国产算力需求侧呈现多维度共振：海外云厂商资本开支已进入新一轮上行 周期，其算力迭代节奏同步牵引全球设备需求，为我国云计算投资提供周 期性机遇；美国对华芯片管制不断升级，从硬件延伸至软件、人才乃至云 服务，形成全链条封锁，反而倒逼国内形成以"自主可控"为核心的政策 与产业共识；中国自上而下全面布局，以"东数西算"国家工程为牵引， 结合地方"算力券""模型券"等创新激励，构建起覆盖战略规划、基础 设施建设与场景开放的完整政策体系；同时，以 DeepSeek-V2 为代表的轻 量化模型技术取得突破，通过稀疏混合专家架构（MoE）等创新，在保持高 性能的同时大幅降低训练与推理的算力负担，为国产芯片切入主流 AI 应 用扫清了关键的性能门槛。 自主技术突破，夯实国产算力供给底座 面对强劲的需求牵引，国产算力供给侧在技术攻坚与生态构建上取得系列实质 性突破。在硬件层面，国内先进制程在持续追赶的同时，以 Chiplet（芯粒） 技术路径实现"制程混搭"，将计算核心与 I/O 等模块解耦，有 ...

Core Viewpoint - The article discusses the recent shift in U.S. technology policy regarding the export of AI chips to China, highlighting the implications of this decision for both American companies and the global AI chip industry. It emphasizes the complex interplay of business interests, national security, and the evolving landscape of China's domestic chip industry. Group 1: Policy Shift and Its Implications - On December 8, 2025, President Trump announced the approval for NVIDIA to export H200 AI chips to China, with the U.S. government taking a 25% cut from each sale, marking a significant policy reversal from previous export bans [1][2] - The decision followed a closed-door meeting between NVIDIA CEO Jensen Huang and U.S. government officials, where Huang sought to restore chip exports to mitigate NVIDIA's declining sales in China, which had resulted in a write-down of approximately $5.5 billion [4][5] - The U.S. government aims to balance maintaining its chip market position while ensuring technological superiority and extracting financial benefits from the deal [5][6] Group 2: Economic Analysis of the Deal - Industry estimates suggest NVIDIA could export between $2 billion to $5 billion worth of chips quarterly to China, meaning the U.S. government could earn $500 million per quarter, totaling $2 billion annually without any investment in R&D or manufacturing [6][7] - This arrangement is characterized as "rent-seeking," where the U.S. government collects a "protection fee" from NVIDIA, which still finds the deal beneficial compared to the zero revenue it faced under the export ban [7][8] Group 3: China's Response and Market Dynamics - Despite the U.S. expectations, the Chinese market's response to the H200 chips has been more cautious than anticipated, with concerns over reliability and cost-effectiveness [29][30] - Factors influencing China's reluctance include a broken trust in U.S. supply chains, the high cost of H200 due to the 25% fee, and the potential security risks associated with U.S. technology [31][33][34] - The Chinese government is actively promoting domestic chip development, which further limits the market for imported chips [36] Group 4: The Rise of China's AI Chip Industry - Reports predict that by 2026, Huawei will capture 50% of the Chinese AI chip market, while NVIDIA's share could plummet from 39% to just 8% [13][15] - The growth of domestic chip manufacturers like Huawei, Cambricon, and Baidu is accelerating, with significant advancements in AI chip technology and production capacity [16][18][19] - The rapid development of China's AI chip industry is seen as a response to U.S. export restrictions, with the potential to reshape the global semiconductor landscape [19][24] Group 5: Long-term Consequences and Strategic Miscalculations - The U.S. government's strategy of using national security as a pretext for economic gain may backfire, accelerating China's technological independence and reducing global reliance on U.S. technology [20][40] - The contrasting approaches of the Trump and Biden administrations highlight a fundamental tension in U.S. policy towards China, with implications for international trust in U.S. technology [22][40] - Ultimately, the article suggests that true victory lies in fostering domestic innovation rather than relying on outdated technology imports, emphasizing the long-term nature of technological competition [44]

Core Viewpoint - The domestic AI industry is experiencing significant growth, driven by strong policy support, technological advancements, and surging market demand, particularly highlighted by the performance of the domestic AI-focused ETF and its constituent stocks [1][4][5]. Group 1: Market Performance - The domestic AI-focused ETF (589520) saw an intraday price increase of nearly 3%, currently up 1.81%, marking a three-day consecutive rise [1]. - Key constituent stocks such as Chipone Technology surged over 15%, while Cambricon and Yuke Tech rose by more than 4% and 3% respectively [1]. - The ETF's trading threshold was reduced from approximately 120 yuan to about 60 yuan following a 1:2 share split, making it more accessible for investors [5]. Group 2: Industry Trends - Major global tech companies, including Oracle, Microsoft, Google, and Meta, are significantly increasing their capital expenditures on AI infrastructure, indicating a robust demand for high-end chips and related hardware [3]. - The Chinese government's policies are strongly supporting the integration of AI across key sectors, with initiatives like the "Artificial Intelligence+" action plan aiming for deep integration by 2027 [4]. Group 3: Technological Advancements - Domestic companies are making rapid technological breakthroughs, with Huawei and Alibaba Cloud demonstrating advancements in large-scale computing capabilities [4]. - The processing capacity in China has dramatically increased, with daily token processing volume soaring from 100 billion to 30 trillion within a year and a half [4]. Group 4: Investment Opportunities - The ETF is positioned to benefit from the ongoing domestic AI industry growth, with a focus on companies that are well-placed in the AI supply chain [6][7]. - The ETF's top ten holdings account for over 71.66% of its weight, with semiconductors being the largest sector, representing over 54.1% [7].

Group 1 - The U.S. government has imposed export controls on NVIDIA's H20 chip, requiring licenses for sales to China, which indicates a significant tightening of trade regulations in the AI chip sector [2][6] - NVIDIA's CEO Jensen Huang visited China again, expressing a desire to continue collaboration, highlighting the impact of these export restrictions on the industry [1][2] - The restrictions on H20 and similar AI chips from AMD and Intel create a substantial opportunity for domestic Chinese AI chip manufacturers to capture market share previously held by NVIDIA [6][14] Group 2 - Huawei's Ascend 910C chip is currently the most prominent domestic alternative, with its latest generation products being integral to China's AI infrastructure [7][14] - The Ascend 910C chip achieves a computing power of 800 TFLOP/s (FP16) and a memory bandwidth of 3.2 TB/s, which is approximately 80% of NVIDIA's H100 performance [7][10] - Despite its advantages, the Ascend 910C faces challenges such as increased power consumption and potential interconnect bottlenecks, which could hinder its efficiency in large-scale AI training tasks [9][11] Group 3 - The production of the Ascend 910C chip relies heavily on TSMC for manufacturing, as domestic foundries like SMIC struggle with yield rates and production capacity [12][13] - The supply chain for critical components, such as HBM memory, is also a concern, as it involves complex logistics and potential legal issues [13][14] - A diverse ecosystem of domestic AI chip companies, including Alibaba, Baidu, and Tencent, is essential for reducing risks and ensuring stability in the AI chip market [14][15] Group 4 - Companies like Cambrian and Moore Threads are emerging as significant players in the AI chip market, with innovative designs and capabilities that can compete with NVIDIA's offerings [15][16] - The development of a unified software platform by Moore Threads aims to facilitate the transition for users from NVIDIA's CUDA to their own architecture, enhancing user experience and performance [16] - The collaboration among various domestic firms is crucial for building a resilient and self-sufficient AI chip ecosystem in China, moving away from reliance on uncertain supply chains [17]