点击上方"公众号"可订阅哦！ 兴业证券计算机小组 蒋佳霖/孙乾/杨本鸿/陈鑫/张旭光/杨海盟/桂杨 本周观点聚焦 1、本周 观 点： 财报披露冲刺期将至，底部加仓质优龙头 2、深度跟 踪 ： 信创复盘研究：观往知来，超额成长可期 周观点 财报披露冲刺期将至，底部加仓质优龙头 财报披露冲刺窗口将至，建议底部加仓。 截至到 04/19 ，已正式披露 2024 年年报、 2025 年一季报的计算机公司数量分别为 133 家、 12 家，占比为 40% 、 4% ， 距离财报季结束还有 8 个工作日，财报将密集披露。伴随着财报的披露，以及国际形势的变化，板块近期持续调整，但业绩趋势及产业变化偏正向的格局并未变化。同 时，对于市场关心的月历效应，根据统计，自 2010 年以来，计算机指数在 4/5/6 月上涨概率分别为 20% 、 53% 、 60% 。综上，继续建议底部加仓。 MCP 加速 AI 应用落地，积极关注其进展。 MCP （ Model Context Protocol ）通过标准化接口实现 AI 模型与外部工具 / 数据的无缝交互，有助于 Agent 应用快速搭 建。近期，阿里云百炼、腾讯云、百度地图等均 ...

Core Insights - The Model Context Protocol (MCP) has significantly monopolized the middle layer of Agentic AI, with its usage growing rapidly since its open-source release in November last year [4][5][6] - MCP is likened to a USB-C port, aiming to become a standardized interface for AI applications, facilitating seamless integration and interaction with various data sources and tools [3][21] - The emergence of the MCP ecosystem is evident, with a variety of MCP Clients and Servers, as well as a marketplace and infrastructure developing around it [7][8] Insight 01: MCP's Dominance - MCP has established itself as a dominant middle layer for Agentic AI, allowing systems to provide contextual information to AI models and enabling integration across various scenarios [4][5] - The protocol simplifies the integration process for developers, enhancing the user experience of LLMs by providing a unified way to access data sources [4][5] Insight 02: MCP Ecosystem Development - The MCP ecosystem is rapidly expanding, with a rich variety of MCP Clients and Servers emerging, alongside dedicated marketplaces and infrastructure products [7][8] - MCP Clients can seamlessly connect to any MCP Server to obtain context, while MCP Servers allow tool and API developers to easily gain user adoption [8][9] Insight 03: MCP as a Standardized Interface - MCP serves as a standardized interface between LLMs and data sources, facilitating the transformation of various data types into a unified format for AI applications [21][22] - The protocol redistributes the workload of data transformation, allowing independent developers to create effective connectors for various applications [22] Insight 04: Maximizing Context Layer Effectiveness - To fully leverage AI Agents, three core elements are essential: rich context, a complete tool usage environment, and iterative memory [24] - MCP enhances the effectiveness of the Context Layer by enabling community-driven development and optimization, which is crucial for high-quality AI agents [25] Insight 05: MCP as a Comprehensive Solution - MCP consolidates various existing middle-layer products into a more lightweight and open foundational protocol, impacting competitors like OpenAI's Function Call and LangChain [29][30] - The protocol's modularity and ecological potential are highlighted, allowing for broader adoption and integration across different platforms [31] Insight 06: MCP's Role in Agentic AI - MCP is positioned as an open protocol that facilitates access to context and tools for users who do not have control over the underlying systems [32] - The flexibility of MCP allows it to serve as a robust solution for developers looking to integrate various data sources and tools into their applications [32] Insight 07: Entrepreneurial Opportunities in the MCP Ecosystem - The MCP ecosystem presents three main entrepreneurial opportunities: Agent OS, MCP Infrastructure, and MCP Marketplace [33][35] - The development of scalable MCP Servers and a marketplace for discovering and installing MCP Servers are key areas for growth and innovation [39][40]