行 业 研 究 / 行 业 深 度 相关研究 《Open Claw 会引领 Agent 元年吗？— — GenAI 系列报告之 72》 2026/03/17 《从智谱 IPO 看当前大模型进展和投资机 遇——GenAI 系列报告之 71》 2026/03/09 证券分析师 林起贤 A0230519060002 linqx@swsresearch.com 任梦妮 A0230521100005 renmn@swsresearch.com 黄俊儒 A0230525070008 huangjr@swsresearch.com 洪依真 A0230519060003 hongyz@swsresearch.com 陈晴华 A0230525100001 chenqh@swsresearch.com 黄忠煌 A0230519110001 huangzh@swsresearch.com 研究支持 陈悦 A0230524100003 chenyue@swsresearch.com 联系人 任梦妮 A0230521100005 renmn@swsresearch.com 2026 年 03 月 24 日 从 MiniMax 看国产大 ...

行 业 研 究 / 行 业 深 度 相关研究 《Open Claw 会引领 Agent 元年吗？— — GenAI 系列报告之 72》 2026/03/17 《从智谱 IPO 看当前大模型进展和投资机 遇——GenAI 系列报告之 71》 2026/03/09 证券分析师 林起贤 A0230519060002 linqx@swsresearch.com 任梦妮 A0230521100005 renmn@swsresearch.com 黄俊儒 A0230525070008 huangjr@swsresearch.com 洪依真 A0230519060003 hongyz@swsresearch.com 陈晴华 A0230525100001 chenqh@swsresearch.com 黄忠煌 A0230519110001 huangzh@swsresearch.com 研究支持 陈悦 A0230524100003 chenyue@swsresearch.com 联系人 任梦妮 A0230521100005 renmn@swsresearch.com 2026 年 03 月 24 日 从 MiniMax 看国产大 ...

开源生态的协同发展，进一步巩固了成本与技术的双重优势。国产开源大模型企业摒弃"闭门造车"的模 式，积极推动生态共建，Kimi K2.5与Cursor的授权合作，正是开源生态协同的生动体现。这种合作不仅 实现了技术的高效落地，更通过资源共享、优势互补，进一步摊薄研发与运营成本，推动技术迭代速度 持续加快。 国产开源大模型能够走到当前的位置，不是单一维度的突破，而是成本控制、技术创新与生态共建的系 统性成功。当然，创新之路上仍需警惕挑战，如核心技术的深度原创、开源生态的规范发展等，都是未 来需持续发力的方向。但不可否认，成本筑基、技术赋能的发展路径，已让国产开源大模型在全球竞争 中占据主动。 责编：叶舒筠 校对： 许欣 MiniMax M2.5连续五周位居全球大模型调用量榜首，月之暗面Kimi K2.5成为知名编程平台Cursor的核 心支撑。 近日的这两则新闻印证了国产开源大模型已从技术跟跑迈入并跑领跑行列。在全球AI竞争日趋激烈的 当下，中国企业凭借精准的成本控制与持续的技术迭代，走出了一条独具特色的创新之路。 成本优势是国产开源大模型突围的基石，也是赢得全球市场青睐的核心底气。国产大模型的成本优势， 源于能 ...

Oracle 业绩验证 AI 算力需求强劲 [Table_Page] 跟踪分析|传媒 证券研究报告 [Table_Title] 互联网传媒行业·AI 周专题 [Table_Summary] 核心观点： [Table_Grade] 行业评级 买入 前次评级 买入 报告日期 2026-03-17 [Table_PicQuote] 相对市场表现 [分析师： Table_Author]旷实 SAC 执证号：S0260517030002 SFC CE No. BNV294 010-59136610 kuangshi@gf.com.cn 分析师： 廖志国 SAC 执证号：S0260525060001 021-38003665 liaozhiguo@gf.com.cn 分析师： 宾特丽亚 SAC 执证号：S0260526010003 021-38003624 binteliya@gf.com.cn 请注意，廖志国,宾特丽亚并非香港证券及期货事务监察委 -20% -7% 6% 18% 31% 44% 03/25 05/25 08/25 10/25 12/25 03/26 传媒 沪深300 员会的注册持牌人，不可在香港从事受监 ...

Investment Rating - The report maintains an "Accumulate" rating for the computer industry, indicating a positive outlook for the sector in the next 6 to 12 months [1]. Core Insights - The integration of computing power and electricity (算电协同) is being driven by supportive national policies, which have been progressively implemented since 2021. This integration aims to create a smart closed-loop system where computing power drives electricity demand and vice versa [3][9]. - The reduction in electricity costs, particularly through green energy hubs, is enhancing the competitiveness of Chinese token exports. As of February 2026, Chinese models accounted for 61% of token consumption, surpassing the U.S. for the first time, with a year-on-year increase of 421% in token consumption [3][20]. - The report outlines three main technical routes for computing power and electricity integration: physical direct supply, virtual direct supply, and cluster direct supply, each with distinct operational models and benefits [3][20]. - Various business models are emerging within this integration framework, including Energy Management Contracts (EMC), Power Purchase Agreements (PPA), and virtual power plant participation in electricity markets [3][20]. - The report identifies four categories of investment opportunities within the computing power and electricity integration sector: traditional power transformation companies, green energy operators, scheduling software service providers, and leading power engineering firms [3][32]. Summary by Sections Policy Support and Industry Upgrade - The integration of computing power and electricity is supported by a series of national policies aimed at creating a collaborative framework for energy and computing resources [9][10]. - The government has outlined specific goals for the development of computing power infrastructure, emphasizing the importance of green energy and efficiency [10][11]. Development of Computing Power and Energy Industry - The report highlights the significant reduction in electricity costs due to the establishment of green energy hubs, which can lower operational costs for data centers [20]. - The competitive advantage of Chinese token exports is attributed to lower electricity costs, with a stark difference in operational costs compared to U.S. models [20]. Technical and Business Models - The report details the three technical models for integrating computing power and electricity, emphasizing their operational efficiencies and cost benefits [20][25]. - Various business models are being adopted, including EMC and PPA, which facilitate long-term partnerships and cost savings for data centers [20][31]. Investment Recommendations - The report suggests that traditional power transformation companies, green energy operators, scheduling software service providers, and leading power engineering firms represent key investment opportunities in the computing power and electricity integration sector [32][33][34].

Core Insights - The article highlights that domestic AI models in China have surpassed U.S. models in weekly usage for two consecutive weeks, indicating a significant shift in the AI landscape [5][6]. Group 1: Domestic Model Performance - The weekly usage of domestic AI models reached approximately 4.69 trillion tokens, an increase of 11.82% from the previous week [6]. - The top three domestic models by usage are MiniMax M2.5 (1.75T tokens), Step 3.5 Flash (1.34T tokens), and DeepSeek V3.2 (1.04T tokens) [5][6]. - In contrast, U.S. AI models had a weekly usage of 3.294 trillion tokens, which represents a decline of 9.33% [6]. Group 2: Emergence of New Models - The newly launched Hunter Alpha model, with 1 trillion parameters and support for 1 million tokens context, has gained attention for its capabilities in long-term planning and complex reasoning [6][7]. - Hunter Alpha topped the daily ranking on OpenRouter shortly after its release, alongside another model, Healer Alpha, which also made it to the top ten [7]. Group 3: Market Dynamics and Pricing - The increasing demand for domestic models is driven by the rise of intelligent agent scenarios, which require high token consumption, making cost-effective domestic models appealing to overseas developers [7]. - For instance, MiniMax M2.5 offers a competitive pricing structure at $0.3 per million tokens for input and $1.1 for output, significantly lower than the prices of U.S. models like Claude Opus 4.6 [7]. Group 4: Commercialization Challenges - While domestic models are gaining traction in international markets, U.S. models are focusing on pragmatic commercialization, such as discontinuing lower-priced versions and tightening usage limits [8]. - The departure of a key figure from Alibaba's AI division highlights the tension between revenue pressures and open-source strategies [8]. - MiniMax's first financial report indicates a revenue of approximately $79.04 million for 2025, a year-on-year increase of 159%, but also reveals a significant loss of $1.87 billion, up 302% year-on-year [8].

Investment Rating - The report maintains a "Buy" rating for the electric power equipment sector [7][22]. Core Viewpoints - The explosive growth of the AI Agent application OpenClaw is becoming a significant catalyst in the global AI industry chain, with a total token call volume reaching 10.4 trillion tokens in March 2026, marking a 30% week-on-week increase [6][15]. - The report highlights the potential for the Chinese wind power industry chain to expand its overseas market share due to cost and delivery advantages, particularly in light of the UK's recent policy changes that eliminate import tariffs on wind power components [20][21]. - The OpenClaw ecosystem is expected to create a transmission path from application growth to increased token demand, leading to an expansion in computing power, data center construction, and upgrades in electrical infrastructure [18]. Summary by Sections Investment Viewpoints - The report suggests focusing on companies such as Goldwind Technology, Daikin Heavy Industries, and others within the wind power sector due to their competitive advantages in cost and delivery [21]. - The electric power equipment sector is expected to benefit from the growth in AI applications and the wind power industry, maintaining a positive outlook for long-term investments [7][22]. Industry Dynamics - The report notes that the AI industry is transitioning from single-round reasoning to multi-round automated reasoning, significantly increasing the demand for computing power [17]. - The UK has announced the cancellation of import tariffs on 33 wind power components, which is expected to accelerate offshore wind installations and increase equipment import demand [20]. - The report emphasizes the importance of regulatory support for the healthy expansion of the OpenClaw ecosystem, which is crucial for the long-term stability of the AI Agent industry [17]. Key Companies and Profit Forecasts - The report provides profit forecasts for several companies, indicating expected earnings per share (EPS) and price-to-earnings (PE) ratios for 2024 to 2026, with specific recommendations for investment [10][23].

Investment Rating - The report maintains a "Buy" rating for the electric power equipment sector [7][22]. Core Insights - The explosive growth of the AI Agent application OpenClaw is becoming a significant catalyst in the global AI industry chain, with a total token call volume reaching 10.4 trillion tokens in March 2026, marking a 30% week-on-week increase [6][15]. - The report highlights the potential for the Chinese wind power industry chain to expand its overseas market share due to cost and delivery advantages, particularly in light of the UK's recent policy changes that eliminate import tariffs on wind power components [20][21]. Summary by Sections Investment Viewpoint - The OpenClaw ecosystem is driving a surge in token calls, which is expected to enhance the demand for domestic computing power, IDC, and related electric power equipment industries in the medium to long term [15][18]. - The UK’s cancellation of import tariffs on wind power components is anticipated to accelerate offshore wind installations, with an estimated investment of £22 billion [20][21]. Industry Dynamics - The electric power equipment sector has shown strong performance, with a 55.8% increase over the past 12 months, outperforming the Shanghai Composite Index [3][4]. - The report emphasizes the importance of the AI Agent paradigm in increasing inference computing power demand, which will benefit the domestic computing power industry chain and accelerate the construction of AI data centers [18][19]. Key Companies and Profit Forecasts - The report identifies key companies to watch, including Goldwind Technology, Daikin Heavy Industries, and others, with specific earnings per share (EPS) and price-to-earnings (PE) ratios provided for 2024 to 2026 [10][23]. - For instance, Goldwind Technology is projected to have an EPS of 0.44 in 2024, increasing to 0.78 by 2026, with a PE ratio decreasing from 71.43 to 40.29 over the same period [10][23].

Investment Rating - The industry investment rating is "Positive" (maintained) [2] Core Insights - Domestic major companies are seizing market opportunities, with the "lobster craze" igniting growth in domestic AI models [6][15] - OpenAI's acquisition of OpenClaw marks a significant step towards the commercialization of AI [21] - The report highlights a surge in token consumption for domestic models, indicating increased user engagement [16] Summary by Sections Domestic Market - Major companies like Tencent, ByteDance, and Zhiyun are launching various AI applications, integrating with OpenClaw to enhance their offerings [6][15] - OpenClaw's usage has surged, with MiniMax M2.5 and Step 3.5 Flash leading in token consumption, accounting for approximately 1/6 of the total weekly token usage on the platform [16][20] International Market - OpenAI's acquisition of OpenClaw signifies a shift towards commercializing AI technologies, with other tech giants like NVIDIA and Meta also investing in AI platforms [21][22] - Despite the excitement in China, the enthusiasm for AI in Europe and the US remains limited, primarily confined to technical circles [22] Policy and Support - The Chinese government is actively supporting AI initiatives, with local policies providing substantial financial backing for qualifying projects [7][20] - Various regions are offering significant subsidies and support measures to promote AI development [20] Investment Recommendations - In the internet sector, focus on AI commercialization and application expansion, recommending stocks like Alibaba-W, Pinduoduo, and Baidu Group-SW, with Tencent Holdings as a beneficiary [9][31] - In the computing sector, there is an anticipated increase in AI demand, with recommendations for companies benefiting from IT spending, including Kingdee International and Baidu Group-SW [9][31] - In the automotive and autonomous driving sector, the approval of domestic L3 pilot licenses is expected to accelerate the commercialization of advanced driving technologies, with recommended stocks including XPeng Motors-W and Tesla [9][31]

Investment Rating - The industry investment rating is "Positive" (maintained) [2] Core Insights - Domestic major companies are seizing market opportunities, with the "lobster craze" igniting growth in domestic AI models [6][15] - OpenAI's acquisition of OpenClaw marks a significant step towards the commercialization of AI [21] - The report highlights a surge in token consumption for domestic models, indicating increased user engagement and model usage [16] Summary by Sections Domestic Market - Major companies like Tencent, ByteDance, and Zhizhu are launching various AI applications, integrating with OpenClaw to enhance their offerings [6][15] - OpenClaw's usage has significantly increased, with MiniMax M2.5 and Step 3.5 Flash accounting for approximately 1/6 of the total token consumption on the platform [16][20] International Market - OpenAI's acquisition of OpenClaw signifies a shift towards commercializing AI technologies, with other tech giants like NVIDIA and Meta also investing in AI platforms [21][22] - Despite the excitement in China, the enthusiasm for AI models in Europe and the US remains limited, primarily confined to technical circles [22] Weekly Data Update - The Hang Seng Index experienced a decline of 1.13%, with small-cap and technology stocks showing resilience [23][26] - The net inflow of capital in the Hong Kong Stock Connect reached 46.46 billion, indicating a positive shift in investor sentiment [24][30] Investment Recommendations - In the internet sector, focus on AI commercialization and application expansion, recommending stocks like Alibaba-W, Pinduoduo, and Baidu Group-SW, with Tencent Holdings as a beneficiary [9][31] - In the computer sector, highlight opportunities arising from increased IT spending by state-owned enterprises, recommending companies like Kingdee International and BaiRong Cloud [9][31] - In the automotive and autonomous driving sector, the approval of domestic L3 pilot licenses is expected to accelerate the commercialization of advanced driving technologies, with recommended stocks including XPeng Motors-W and Xiaomi Group-W [9][31]