华西证券表示，近期中国AI大模型迎来"春节档"爆发，技术趋势已从通用聊天工具全面转向垂直生产力工具与真实Agent落地，重点体现在工业 级的视频生成（如Seedance 2.0的多模态叙事与音画同步）、工程级编程（如智谱GLM-5的跨文件重构与系统工程能力）以及消费办公场景。目 前AI发展阶段仍处于关键加速期，伴随相关算力芯片的供应体系逐渐丰富，应用发展对于token需求量仍在加速增加，相关底层算力基建仍然还 在扩张期。 或被海外AI情绪错杀，重点布局国产AI产业链的科创人工智能ETF（589520）连跌两日，场内价格现跌1.76%，回踩10日均线。上交所数据显 示，该ETF近2个交易日连续吸金3099万元，反映资金看好国产AI后市表现，逢跌进场埋伏！ 成份股方面，道通科技涨超2%，中科星图、合合信息涨逾1%，寒武纪、安路科技等个股飘红。另一方面，星环科技跌超9%，芯原股份跌逾 5%，跌幅居前，拖累指数表现。 【国产替代之光，科创自立自强】 科创人工智能ETF华宝（589520）及其联接基金（联接A：024560，联接C：024561）重点布局国产AI产业链，成份股囊括国产GPU龙头（如寒 武纪），国产ASI ...

Group 1 - The core viewpoint is that the domestic AI industry is experiencing significant investment interest, as evidenced by the recent performance of the Science and Technology Innovation Artificial Intelligence ETF (589520), which has seen a decline of 1.76% over two days while attracting 30.99 million yuan in inflows, indicating confidence in the future of domestic AI [1][8] - The ETF's composition includes leading domestic companies in the AI sector, such as Cambricon Technologies and Chipone Technology, with a strong focus on semiconductor and software industries, which are expected to benefit from the AI application surge [4][11] - Recent reports highlight a surge in demand for domestic AI models, with predictions of explosive growth in token usage by 2026, reflecting an exponential increase in AI inference demand and the potential for domestic computing power to dominate the infrastructure layer [10][11] Group 2 - The ETF is designed as an efficient tool for investors to gain exposure to the domestic computing power sector, with a significant portion of its holdings in semiconductor stocks, which account for nearly half of the ETF's weight [4][11] - The report from Huaxi Securities indicates that the current phase of AI development is critical, with advancements in industrial-grade video generation and programming tools, suggesting a shift from general chat tools to specialized productivity applications [10] - The fee structure for the ETF and its linked funds is outlined, with no sales service fees and specific commission rates for subscription and redemption, making it accessible for investors [6][13]

| 代码 | 名称 | 现价 | 涨跌 | 涨跌幅 ▲ | | --- | --- | --- | --- | --- | | 588930 | 科创人工智能ETF银华 | 1.724 | -0.038 | -2.16% | | 589230 | 科创人工智能ETF南方 | 0.956 | -0.021 | -2.15% | | 588760 | 科创人工智能ETF广发 | 0.837 | -0.019 | -2.22% | | 589090 | 科创AIETF鹏华 | 1.178 | -0.025 | -2.08% | | 589520 | 科创人工智能ETF华宝 | 0.667 | -0.014 | -2.06% | | 588790 | 科创AIETF博时 | 0.884 | -0.018 | -2.00% | | 588730 | 科创人工智能ETF易方达 | 1.630 | -0.033 | -1.98% | | 589010 | 科创人工智能ETF华夏 | 1.560 | -0.031 | -1.95% | | 589380 | 科创AIETF富国 | 1.564 | -0.030 | -1.88 ...

截至2026年2月25日 09:45，中证人工智能产业指数(931071)下跌0.05%。成分股方面涨跌互现，星宸科 技领涨2.16%，均胜电子上涨2.09%，慧博云通上涨2.08%；芯原股份领跌，澜起科技、星环科技跟跌。 人工智能ETF(515980)盘中成交4378.14万元。拉长时间看，截至2026年2月24日，人工智能ETF近一年累 计上涨56.44%。 中信证券研报称，2026春节期间，国产大模型token调用量井喷，截至2月22日此前一周的全球大模型 token用量中，前三甲均为国产大模型。中信证券认为，token的爆发式增长，本质上反映出AI推理需求 的指数级扩容，而国产算力凭借着成本优势及不断完善的生态，有望在基础设施层逐步占据主导。 华富人工智能ETF（515980）跟踪中证人工智能产业指数，依托标的指数三层闭环编制逻辑+季度调仓 紧跟产业趋势的优势，是布局A股AI的优质小宽基标的。在同类AI相关指数中具备较高的AI应用含量， 将40%的权重聚焦于应用端，同时也保持了60%的算力基础设施这一产业共识核心。结合春季行情日历 效应，当前市场震荡回调中，投资者可密切关注人工智能ETF，把握AI全产 ...

数据显示，江信基金旗下1只基金重仓芯原股份。江信瑞福A（002630）四季度持有股数200股，占基金 净值比例为2.43%，位居第四大重仓股。根据测算，今日浮亏损失约3044元。 江信瑞福A（002630）成立日期2017年2月17日，最新规模51.31万。今年以来收益7.35%，同类排名 3044/8889；近一年收益17.74%，同类排名4915/8136；成立以来收益55.12%。 江信瑞福A（002630）基金经理为高鹏飞。 截至发稿，高鹏飞累计任职时间5年124天，现任基金资产总规模1672.91万元，任职期间最佳基金回报 29.45%， 任职期间最差基金回报16.16%。 2月25日，芯原股份跌5.91%，截至发稿，报242.31元/股，成交9.55亿元，换手率0.74%，总市值1274.35 亿元。 资料显示，芯原微电子(上海)股份有限公司位于中国(上海)自由贸易试验区春晓路289号张江大厦20A， 成立日期2001年8月21日，上市日期2020年8月18日，公司主营业务涉及依托自主半导体IP,为客户提供 平台化、全方位、一站式芯片定制服务和半导体IP授权服务。主营业务收入构成为：芯片量产业务收 ...

截至2026年2月25日 09:32，中证人工智能主题指数(930713)下跌0.30%。成分股方面涨跌互现，星宸科 技领涨4.17%，协创数据上涨1.23%，拓维信息上涨1.21%；芯原股份领跌4.09%，昆仑万维（维权）下 跌2.75%，北京君正下跌2.47%。AI人工智能ETF(512930)下跌0.40%，最新报价2.27元。 资金流入方面，AI人工智能ETF最新资金净流入3194.73万元。拉长时间看，近4个交易日内有3日资金 净流入，合计"吸金"1.25亿元，日均净流入达3118.97万元。（数据来源：Wind） 上个马年沪指涨近60%！新年新福利来了，炒股排面要拉满，新号启幸运>> 消息面上，全球最大AI开源社区Hugging Face公布新一期开源大模型榜单，阿里最新开源的原生多模态 模型千问3.5登顶全球榜首。在最新一期Hugging Face模型榜单前十中，有8款来自中国团队，涵盖阿里 巴巴、智谱、MiniMax、月之暗面等多家机构。 中信证券研报指出，2026春节期间，国产大模型token调用量井喷，截至2月22日此前一周的全球大模型 token用量中，前三甲均为国产大模型。token的 ...

Group 1 - The Integrated Circuit ETF (159546) opened with a gain of 0.29%, priced at 2.063 yuan [1] - Major holdings in the ETF include: Cambrian (up 0.79%), SMIC (up 1.02%), Haiguang Information (up 0.75%), and others with varying performance [1] - The ETF's performance benchmark is the CSI All-Share Integrated Circuit Index return, managed by Guotai Fund Management Co., Ltd., with a return of 105.49% since its establishment on October 11, 2023, and a one-month return of -4.74% [1]

Market Performance - Historically, A-shares perform well in the two weeks leading up to the Two Sessions and slightly worse during the sessions, with an increased probability of rising post-sessions [2][9] - The small-cap style dominates both before and after the Two Sessions, with the CSI 500 and CSI 1000 indices benefiting from expectations of stable growth policies and industrial policies [2][11] - After the Two Sessions, cyclical sectors are expected to perform better, with higher average returns [2][11] Monetary Policy and Interest Rates - The central bank's net injection in the open market was 23,969 billion yuan last week, with a future injection of 22,524 billion yuan in reverse repos and 3,000 billion yuan in MLF [2][18] - Money market rates are declining, with the R007 and DR007 rates down by 18.4 basis points and 14.0 basis points respectively [2][18] Fund Supply and Demand - In the last week, there was a slight net outflow of funds in the secondary market, with a net sell-off of 830.2 billion yuan in financing [3][23] - The issuance of public funds increased by 340.3 billion units, while ETF saw a net outflow of 482.5 billion yuan [3][23] Market Sentiment - Financing trading activity weakened, with a decrease in the proportion of financing transactions in A-share turnover to 8.6% [3][28] - The VIX index fell, indicating improved risk appetite in overseas markets [3][33] Industry Preferences - The only sectors with net inflows were media and building materials, while significant outflows were seen in sectors like electronics and power equipment [3][35] - The highest net inflow was in the media sector (+15.1 billion yuan), while the largest net outflow was in power equipment (-97.3 billion yuan) [3][36] Industry Performance Post-Two Sessions - Post-Two Sessions, industries such as real estate, building materials, and consumer electronics are expected to see higher probabilities of price increases as stable growth policies are confirmed [2][15]

Core Insights - The report from Kaiyuan Securities highlights significant updates in domestic and international AI models during the Spring Festival, with a recommendation to focus on sectors benefiting from increased AI computing capital expenditure and domestic supply chain control [1][2] Market Review - During the Spring Festival period (February 16-20, 2026), major overseas technology indices saw gains, with the Nasdaq and Philadelphia Semiconductor Index both rising by 1.51%. Key tech stocks such as NVIDIA increased by 3.83%, Apple by 3.44%, Google by 2.90%, and Amazon by 5.69%, while AMD fell by 3.46% [2] - AI upstream semiconductor equipment, materials, and storage experienced widespread increases, with ASML up by 4.48%, AMAT by 5.90%, Micron by 4.01%, and SanDisk by 3.74% [2] Industry Updates - Domestic AI models are entering a dense release period, with updates from companies like Qianwen, Seedance, and others. ByteDance has launched several new models, including Seedance 2.0 and Doubao-Seed-2.0 [3] - Google released the Gemini 3.1 Pro model, achieving a score of 77.1% on the ARC-AGI-2 benchmark, with performance improvements over its predecessor [3] - Apple is set to hold a special event on March 4, 2026, to unveil multiple new products, including the iPhone 17E and a budget MacBook [3] Computing Power - NVIDIA's CEO Jensen Huang announced a groundbreaking new chip at the upcoming GTC 2026 conference. NVIDIA and Meta have signed a multi-billion dollar chip procurement agreement, with Meta committing to purchase millions of NVIDIA's latest AI chips [4] - Reports indicate a second round of price increases for electronic fabrics due to rising costs and tight supply, with multiple price hikes occurring from October 2025 to February 2026 [4] - OpenAI is finalizing a new funding round aiming to raise $100 billion, which would value the company at $830 billion, alongside a long-term capital expenditure plan projecting $600 billion in computing power spending by 2030 [4] Storage - Samsung and SK Hynix are accelerating their production expansion due to surging storage demand from AI, with Samsung moving up the production timeline for its P4 factory to Q4 2026 [5] - SK Hynix plans to start trial operations at its Longjin plant earlier than expected, focusing on high-value products like high-performance DRAM and HBM [5] - Micron's CFO confirmed that HBM4 memory is now in mass production, with shipment volumes expected to rise this quarter, achieving a transmission rate of 11 Gbps [5]

Investment Rating - The report maintains a "Recommended" investment rating for the electronic industry [4][9]. Core Insights - The report highlights the ongoing competition in AI model development, with significant advancements from both international and domestic companies, particularly in computational power [6]. - It emphasizes the importance of computational power as a key investment theme, focusing on overseas computational power, domestic computational power, and storage as three main lines of investment [6][9]. - The report notes that domestic AI models are reaching a dual inflection point in technology and commercialization, with a shift from parameter competition to efficiency and agent-based innovations [6]. Summary by Sections Overseas Computational Power - Google launched its Gemini 3.1 Pro model, achieving a significant increase in reasoning and coding capabilities, with test scores surpassing previous models [6]. - Companies like Lumentum and Coherent saw stock increases of 19.90% and 14.57% respectively, driven by better-than-expected orders [6]. - Nvidia's stock rose by 4.78% as it prepares to unveil a revolutionary AI chip at the upcoming GTC 2026 conference [6]. Domestic Computational Power - Domestic AI model companies performed well during the holiday period, with notable stock increases for companies like Zhizhu and MiniMax [6]. - The report mentions the successful launch of several new AI models by domestic companies, indicating a shift towards efficiency and commercialization [6]. - Key domestic chips have been adapted for new models, with significant support from various domestic computational platforms [6]. Storage Chips - The report notes stock increases for companies like SanDisk and Micron, driven by growing AI demand [6]. - Micron is investing $50 billion to expand its production capacity in response to increasing AI infrastructure needs [6]. - The report highlights the significant price increase for HBM4 chips, reflecting the inflationary effects of AI demand [6].