Through the unified interface standard, the MCP protocol allows different tools and services to be easily connected to AI agents like “plug and play”, greatly improving development efficiency and user experience. This article delves into the background, technical advantages of the MCP protocol, and the strategic layout of domestic giants such as Baidu, Tencent, and Alibaba in this field.
Imagine that your phone needs to be connected to a charger, headphones, and USB flash drive at the same time, but each device needs a different interface, and this “interface anxiety” was once a nightmare for all of us. In the AI era, if each tool (such as payment and navigation) requires an independent “interface” (API), then developers and users will have to repeatedly “change plugs”.
Fortunately, the “universal socket” MCP (Model Context Protocol) in the AI world is unifying interface standards, allowing different functions and tools to be “plug and play”.
For example, a takeaway AI agent can automatically call restaurant data, payment discounts, and logistics interfaces from multiple platforms through the MCP protocol, eliminating the need for developers to connect one by one. For this AI agent, we can understand it as a “super butler”. It not only understands commands, but also actively calls multiple tools to complete tasks.
Suppose you say “go to Hangzhou”, it can automatically plan the itinerary, book tickets, and calculate the budget. And all of this depends on the MCP protocol to break down the barriers between tools.
And when this “universal socket” goes from the laboratory to the world, an “AI arms race” has recently been staged among major domestic manufacturers.
This war without gunpowder smoke may affect the future changes in the pattern of technology giants in the era of AI agents.
1. Big factories grab the beach, and the “universal socket” rewrites the rules of the game
According to Tianyancha comprehensive information and a report released by market research firm MarketsandMarkets in November 2024, the market size of AI agents is expected to grow from $5.1 billion in 2024 to $47.1 billion in 2030, with a compound annual growth rate of 44.8%.
As the “golden key” to open the 100-billion-level market of AI agents, major manufacturers are vying to adapt technology and implement scenarios based on MCP. From aggregating multi-domain resources to developing vertical scenarios, from expanding multi-modal capabilities to building a commercial closed loop, AI agents are upgrading from command response to cross-modal task collaboration under the layout of large factories.
What does a product manager need to do?
In the process of a product from scratch, it is not easy to do a good job in the role of product manager, in addition to the well-known writing requirements, writing requirements, writing requirements, there are many things to do. The product manager is not what you think, but will only ask you for trouble, make a request:
View details >
MCP (Model Context Protocol) was released by Anthropic in November 2024 and defines the standardized way AI models interact with external systems.
▲ Figure: MCP protocol official website – MCP protocol architecture diagram
At its core, this protocol is the Unified Communications Protocol, which provides a standardized interaction framework for AI models and external tools, and supports dynamic discovery of available services.
After the release of the agreement, it attracted great attention at home and abroad.
April 2025 can be described as a key node for domestic technology giants to intensively deploy MCP+AI agents.
Among them, Baidu builds a three-layer system of “model-MCP-application” through the “Qianfan Platform”, and opens search and map capabilities to create “MCP Square”; Tencent integrates plug-ins such as location services and WeChat reading based on the knowledge engine, and supports custom MCP plug-in calls.
Alibaba launched the “Payment MCP Server” to open up the whole link of transactions, Alibaba Cloud Bailian provides development tools such as “MCP Square”, and open-source integrates the Qwen3 model of MCP to strengthen Agent capabilities.
In addition, ByteDance’s internal testing of the general agent product “Buckle Space” explores MCP integration, and there is no large-scale opening capability yet.
According to the comprehensive information of Tianyancha media, the reason why the MCP protocol is popular with manufacturers and developers is that it breaks through the technical bottleneck of the API development model.
In the past, APIs were the main way of interconnection between tools, and developers had to invest a lot of effort in writing adaptation code, which was inefficient and costly.
The emergence of the MCP protocol provides a unified standard, allowing different services and tools to be easily connected to large models and AI agents like “plug and play”.
This solution to the problem of API interface fragmentation not only allows large manufacturers to see opportunities to improve their technical collaboration capabilities, but also provides a driving force for them to compete for opportunities in the ecological layout of AI agents.
Looking at the layout of various major manufacturers, one question is, why have major manufacturers reinvested in even All IN MCP+AI agents? What kind of “treasure” is bet behind this? Behind the seemingly similar strategies, what kind of pattern does it indicate in the future?
As AI agents will become popular at both TO B and TO C, who among the major manufacturers can compete for more market share and occupy a dominant position in the future may be the key to this “competition”.
2. The three-layer ecological chain behind the hurricane
Based on the MCP protocol, the strategic layout of major manufacturers to break down technical barriers and integrate tools and services is not only to seize the opportunity, but also to bet on the future. After all, the impact of the MCP agreement on the industry has only just been revealed.
From the current point of view, there may be a three-layer ecological chain in the field of AI agents based on the blessing of the MCP protocol.
The first layer is the developer’s “cost reduction and efficiency increase” and “spontaneous team”. As a unified standard, MCP provides active technical support for the development of AI agents and solves many pain points that have plagued developers for a long time, especially in terms of development efficiency and tool interoperability.
MCP+ agents, through unified interface standards, make different tools and services easy to access without the need for developers to do additional adaptation work.
For developers, the greater benefit is also the “spontaneous side” of the MCP protocol. In the past, the development of AI applications often required a large team and high capital investment to focus on a certain platform, but based on the MCP protocol, developers used technical merit as the voting criterion, and user choices were influenced by both habits and experience.
The second layer is that when large manufacturers and developers form synergy, the explosion of AI agents can also benefit from ordinary users, and even the popularization and application of MCP protocol means a leap in AI experience for C-end users.
Whether it’s handling daily affairs through AI agents or planning travel schedules through agents, the introduction of MCP protocols makes these services more intelligent, accurate, and efficient.
For example, when users use intelligent itinerary planning tools, they can seamlessly connect large models with Baidu maps, hotel reservation systems, weather forecasts and other services through the MCP protocol, and finally provide users with preferred travel plans, all of which happen between the user’s fingertips, instantaneously, without frequently switching apps or web pages.
▲ Picture: Baidu Maps official website
The third layer is the “competition” of large factories, which ultimately points to the commercialization of the MCP+AI agent ecosystem.
Although AI agents are still in a relatively rudimentary stage of cognitive education, even the industry lacks unified definitions and recognized standards. However, how AI agents connect with the commercialization path is a more imaginative “pyramid” of interests.
Because of this, major manufacturers that actively respond and promote the popularization of AI agents based on the MCP protocol are expected to achieve greater availability of scenarios, services, and products on the basis of the existing AI ecosystem.
This availability is an improvement of experience for users, and once user habits are formed and new interaction methods become mainstream, then the AI agent ecology based on MCP by large manufacturers has the possibility of forming a commercial closed loop.
Of course, commercialization is not only an ecological derivative, but also the core engine driving the implementation of “new infrastructure”, different from the hardware attributes of traditional infrastructure, MCP+AI agent “new infrastructure” is a soft protocol that connects the ecosystem, and its value lies in breaking the silos of data and tools.
As more developers and enterprises join, MCP+AI agents will show their potential and application value in many fields such as life services, Internet industry, and even manufacturing, financial industry, etc., and may even become an emerging force driving the process of intelligence.
However, with the widespread application of the MCP protocol, industry competition will inevitably become more intense. In this “competition”, whoever can make a breakthrough in technology and build an open and interoperable ecosystem will have the potential to occupy an advantageous position in the future arena.
3. The deep game of ability equality
With the popularization of the MCP protocol and the rapid development of the AI agent ecosystem, the competition between major manufacturers in technology and the market is becoming more and more fierce. The future competition is no longer limited to the competition of a single product, but has evolved into a deep game based on “ability equality”.
Although large manufacturers can rely on their existing advantages in the short term, the open nature of MCP+AI agents may dilute the original barriers to differentiation.
This is the difference between the differentiated “open battle” or the “secret battle” of the new scene in this competition, and the latter may become a game that determines who can win in terms of user usage time and function coverage.
This is because differentiation is a static advantage, while scenarioization is a dynamic capability. Whoever can combine a cross-border service such as “takeaway + travel + social” faster will have the potential to define new rules. The MCP protocol is like a “universal socket”, modularizing the core functions of large manufacturers into pluggable services, which can be called and combined by any developer.
The resulting “ability equality” has made differentiated advantages no longer exclusive, and the focus of competition has shifted to how to combine multi-scenario services.
From the perspective of differentiated “open competition”, major manufacturers have used their existing advantages as a springboard to cut into the MCP battlefield, Baidu re-search, map and other tool integration, Alibaba Qiang in the closed loop of transactions, and Tencent has expanded social scenes, and the differentiation path is clear.
The different advantages of large manufacturers are not only the crystallization of years of “hard work” over the years, but also the “nirvana” to attract developers and users in the face of the tide of artificial intelligence.
However, compared with the differentiated “open struggle”, the “secret fight” based on the new scene may be opened with the MCP+AI agent “tearing off” the differentiated label of the big factory.
According to the comprehensive information of Tianyancha media, the “scenario-based” of MCP+AI agents will make competition more and more apparent with the layout of large factories and the entry of massive developers. In this process, the differentiation based on products, applications and functions will gradually “fade” and become a new competitive pattern based on “scenarios” rather than “products”.
Looking further, the application scenarios of AI agents will become more diversified, and the traditional business boundaries are no longer absolute competitive barriers between technology giants.
Especially with the rapid penetration and widespread popularity of MCP+AI agents on the user side, the existing core capabilities of large manufacturers will no longer be limited to the inherent advantages of their own platforms, but developers will decide how to use the tools of large manufacturers, and more flexibly integrate with more tools, functions and capabilities, and then provide them to users.
When a large manufacturer encapsulates core functions as MCP “standardized plugs”, developers can call them freely as if they were plugged into a socket. The exclusive capabilities of large manufacturers have become an open ecological “universal socket”.
Such as this “scenario-based order reconstruction” also indicates that the future MCP+AI agent “arms race” is not only a competition for a single function and application, but also a one-stop solution for developers to build more complex and diversified application capabilities in the form of “packaging”, which is then distributed by large factories and developers to hundreds of millions of users.
This also makes the focus of competition no longer on a single product and function, but on how to win with services and experiences in multi-dimensional scenarios, and then gain a larger market share in different ecosystems.
From differentiated “open struggle” to scenario-based “secret fight”, MCP+AI agents are changing the competitive landscape between large manufacturers. When no longer relying only on traditional technical advantages, but through the construction of agent ecology and scenario-based application, the process of promoting the openness and interoperability of large manufacturers is sought through dynamic balance.
The chemical reaction of MCP+AI agents has just emerged, and the future AI agent ecosystem will become more complex and full of unpredictable possibilities. Therefore, all changes and reshuffles have just begun.
4. Write at the end
MCP+AI agent not only brings about the change of artificial intelligence technology equality and user inclusion, but also its deeper significance is that as the core carrier of artificial intelligence, MCP+AI agent is expected to move towards full popularization and commercial application, thereby promoting a new round of innovation in the artificial intelligence industry.
With the active layout of large manufacturers, AI agents will not only be a tool, but also a personal intelligent assistant, becoming a powerful partner in people’s work and life, and even an “accelerator” for the intelligent process of all walks of life.
When AI agents become “digital organs”, humans can focus on creative decision-making and leave repetitive tasks to agent collaboration, which is the ultimate meaning of this technological revolution.
