Google is transferring the Agent2Agent (A2A) protocol to the Agentic AI Foundation. This brings the standard for communication between AI agents under the same umbrella as MCP, Anthropic’s protocol for connecting agents to tools and data.
A2A is moving to the Agentic AI Foundation (AAIF), the organization within the Linux Foundation that manages open agent infrastructure. The protocol will thus join MCP, as well as projects such as Block’s goose and OpenAI’s AGENTS.md.
The two protocols address different problems. MCP standardizes how agents connect to tools and data sources. A2A governs how agents communicate with one another. Together, according to Google, they form the infrastructure needed for large-scale multi-agent systems. Just last month, the AAIF announced the biggest MCP update to date.
MCP comes from Anthropic, and A2A from Google. Both companies have thus placed their technology under neutral governance. The idea is that, just as common standards contributed to the creation of the modern web and the mobile ecosystem, common standards will be necessary to create an “internet of agents.”
From an internal project to a broad standard
A2A began as an internal Google initiative. The company unveiled the protocol in April 2025 and transferred it to the Linux Foundation in June of that year. At launch, more than fifty partners backed the protocol, including Atlassian, Box, Intuit, LangChain, MongoDB, and Workday, as well as consulting firms such as Accenture, Deloitte, and PwC.
The AAIF operates under a two-tier governance model: a Governing Board responsible for strategy, budget, and membership policy, and a Technical Committee that approves projects and conducts technical reviews. The Linux Foundation provides the legal and operational structure without dictating the technical direction itself.
Google compares this approach to Android. With open, scalable standards and an ecosystem built around them, the company aims to accelerate adoption. Whether this will succeed remains to be seen.