OpenAI announced it will adopt the Model Context Protocol (MCP) developed by rival Anthropic across its entire product lineup.
This move marks a significant shift for OpenAI, a company traditionally relying on its proprietary models. OpenAI‘s decision to adopt a “successful open-source model” has generated curiosity about potential strategic changes.
On Wednesday, Sam Altman, CEO of OpenAI, expressed his excitement on X (formerly Twitter), announcing that MCP support is already available in the Agent SDK and will soon be integrated into the ChatGPT web and mobile apps and through their API.
Mike Krieger, Chief Product Officer at Anthropic, welcomed OpenAI‘s decision to incorporate MCP, highlighting its usefulness in connecting AI models with existing data and software.
Anthropic released MCP as open-source in November of last year. The protocol enables AI models to interact with external software, files, web data, and third-party application data. Developers have praised MCP as a “universal translator” because it supports bidirectional connections.
Recently, Anthropic has shifted its focus from expanding its general-purpose AI chatbot model (B2C) to improving its coding assistant tools (B2B).
Anthropic offers MCP servers across various platforms, including Google Drive, GitHub, Slack, PostgreSQL, Block, Apollo, Replit, Codium, and Sourcegraph.
MCP’s core strength lies in its client-server architecture, which allows bidirectional data connections between AI models and data sources.
Developers can build MCP clients (such as apps, workflows, etc.) that expose their data to the MCP server while connecting to external servers.
Industry analysts view OpenAI‘s adoption of a competitor’s open-source standard as a significant strategic shift. OpenAI has traditionally prided itself on its technological superiority and built its ecosystem.
Experts predict that the importance of open standards will increase as AI agents become more widespread.
AI agents are likely to be implemented as a collection of specialized agents, and no single company is expected to build and operate all of these agents easily. This concept has given rise to terms like “agent orchestration.”
An industry insider commented, “In the future, a company’s competitive advantage will depend on its technological capabilities and how effectively it can utilize other companies’ open-source tools. We can expect the rise of applications like ‘Manus AI,’ which integrate multiple models without relying exclusively on proprietary development.”