几周之前曾被热议的中国资产重估逻辑，还继续有效吗？ 文 丨 陈晶 制图 丨 黄帧昕 编辑 丨 王姗姗 4 月 2 日特朗普签署所谓 "对等关税" 行政令，宣布 美国对所有贸易伙伴加征 10% 的 "最低基准关 税"， 并对部分贸易伙伴征收更高关税，其中给中国的 "对等关税" 税率达到 34%。此后连续三个交易 日（至 4 月 7 日），全球金融市场暴跌，美国股市蒸发了超过 6 万亿美元。港股则在 4 月 7 日抹平了 开年以来所有涨幅，上证指数创下 8 年来最大单日跌幅。 本周，中美关税争端还在继续升级。4 月 8 日，美国针对中国政府的坚决反制，报复性地宣布对中国 输美商品征收 "对等关税" 的税率由 34 % 提高至 84 %。 中国并不示弱，4 月 9 日反制进一步升级 ——对美加征关税税率由 34 % 提高至 84 %。 "这是一个史诗般的经济和市场事件。" 美国著名基金经理比尔·格罗斯（Bill Gross）在上周表示，本 轮美国关税政策调整的影响，形同 1971 年和金本位制的终结但会带来直接的负面后果，因此他呼吁 投资者不应该试图 "接下落的刀"。 反映市场波动率的芝加哥期权交易所 VIX 指数 ...

Core Insights - The rise of domestic open-source large models and intelligent agents has made "AI equality" a hot topic, with widespread adoption across various industries and scenarios [1] - Tencent has launched the "Hunyuan" large model, utilizing the MoE architecture, with flagship model parameters reaching trillions, leading in both general and specialized application capabilities in China [1][2] - Tencent's AI applications, such as the "Yuanbao" product, are gaining traction, with the product now ranking among the top three AI application assistants in China [2] Group 1: AI Development and Applications - Tencent is focusing on creating a usable and iterative AI intelligent system, emphasizing the importance of stable computing power, convenient tool platforms, and authoritative content sources for effective AI service delivery [3] - The company has developed the Tencent Cloud Intelligent Computing Suite, which allows users to set up and start training AI models within one day, achieving a parallel acceleration ratio of 96% and a daily failure rate of only one-third of the industry average [3] - Tencent's TI platform facilitates a one-stop solution for managing, fine-tuning, and deploying various AI models, resulting in a 70% reduction in data labeling costs and a 30% increase in model training efficiency [3] Group 2: Industry Collaboration and Impact - Tencent has collaborated with over 40 domestic robotics companies, with nearly 20 located in Guangdong province, to support the application of robots in various scenarios [3] - The company has successfully implemented its large models across more than 30 industries, including government, retail, finance, industry, healthcare, education, and cultural tourism [5] - Tencent emphasizes the need for collaborative efforts among the industry, government, and itself to effectively realize the potential of large models in different industry scenarios [5]

Group 1 - The China Securities Digital Economy Theme Index (931582) decreased by 1.52% as of March 31, 2025, with mixed performance among constituent stocks [1] - Leading gainers included Huada Jiutian (301269) up 4.43%, Guanglian Da (002410) up 2.28%, and Sanhuan Group (300408) up 2.16%, while leading decliners were Nasda (002180) down 4.51%, Mingzhi Electric (603728) down 4.40%, and Tonghuashun (300033) down 4.32% [1] - The Digital Economy ETF (560800) fell by 1.65%, with the latest price at 0.77 yuan and a trading volume of 10.7362 million yuan [1] Group 2 - The Digital Economy ETF closely tracks the China Securities Digital Economy Theme Index, which selects listed companies in high digitalization sectors to reflect the overall performance of digital economy theme stocks [2] - As of February 28, 2025, the top ten weighted stocks in the index included Dongfang Caifu (300059), SMIC (688981), and Huichuan Technology (300124), collectively accounting for 50.97% of the index [2] Group 3 - The ongoing technological competition among major countries is intensifying, necessitating the localization of AI computing power, supported by policies aimed at increasing the share of self-controlled computing power [1] - The development of leading models like DeepSeek and AIAgent is expected to significantly increase the demand for inference computing power, marking a shift from training-driven to inference-driven demand [1] - Major tech companies are open-sourcing their models, accelerating the democratization of AI and advancing multimodal technology, which presents new development opportunities for AI applications [1]

独立行情中的不确定性 2025 年 03 月 11 日 邮箱：mouyiling@mszq.com 邮箱：meikai@mszq.com 邮箱：fangzhiyong@mszq.com 研究助理：季宏坤 执业证号：S0100124070013 邮箱：jihongkun@mszq.com A 股策略点评 20250311 [Table_Author] 分析师：牟一凌 分析师：梅锴 分析师：方智勇 执业证号：S0100521120002 执业证号：S0100522070001 执业证号：S0100522040003 ➢ 不一样的美股下跌。昨日(20250310)美股整体大幅调整，其中，纳指跌幅创 下 2022 年 10 月以来的新高。实际上，自 2025 年 2 月 20 日以来，美股整体已 经进入下跌趋势，随之而来的是 VIX 指数的持续攀升(图 1)与纳指加速器指标的 持续回落(图 2)。截至 2025 年 3 月 10 日，美股 VIX 指数已经逐步接近 2024 年 8 月的高点，而纳指加速器指标则远低于负两倍标准差水平。从历史上看，每当 纳指加速器指标回落至负一倍标准差以下时，纳指往往会迎来一轮反弹，但 ...

Core Viewpoint - The AI large model sector is experiencing a new wave of price reductions, driven by competition and advancements in algorithms and computing cost control [4][5]. Group 1: Price Reductions and Promotions - DeepSeek has launched a time-limited discount for its API calls, reducing DeepSeek-V3 to 50% of its original price and DeepSeek-R1 to 25% [2]. - Major companies like ByteDance, Alibaba Cloud, Tencent, and iFlytek have announced significant price cuts for their models, with ByteDance's Doubao visual understanding model dropping to 0.003 yuan per thousand tokens, an 85% reduction from the industry average [4]. - Free service strategies are being adopted by several firms, with Baidu's Wenxin Yiyan and OpenAI's ChatGPT offering free access starting April 1 [4]. Group 2: Technological Advancements - NVIDIA has open-sourced a new optimization scheme based on the Blackwell architecture, achieving a 20-fold reduction in single token costs and a 25-fold performance increase compared to the previous H100 model [2][3]. - The new FP4 quantization model reaches 99.8% performance of FP8 models in the MMLU general intelligence benchmark, showcasing significant advancements in model efficiency [3]. Group 3: Market Reactions and Investments - The A-share market has seen a surge in large model concept stocks, with 69 companies experiencing an increase, and a net inflow of 90.02 billion yuan in leveraged funds [5]. - Notable stock increases include Capital Online rising by 13.71% and several others by over 5% [5]. - China Unicom has optimized its "adaptive slow thinking" model for DeepSeek series, saving approximately 30% in inference computation [5]. Group 4: Industry Dynamics - The open-sourcing of DeepSeek's cost-reduction methodology is expected to lower training and inference costs across the industry, stimulating rapid expansion of AI applications [5]. - The competitive landscape is intensifying, with companies focusing on algorithm optimization and iteration as key strategies alongside price reductions [4].