Google's Agent2Agent protocol is moving under the same roof as Anthropic's MCP

Google's Agent2Agent (A2A) protocol — an open standard for letting independent AI agents discover each other and coordinate tasks — is becoming a hosted project of the Agentic AI Foundation (AAIF), according to Axios. A2A had already moved to the Linux Foundation's broader open-governance umbrella in mid-2025; this narrows that further, placing it specifically alongside the Model Context Protocol (MCP), the tool-and-data-connection standard Anthropic donated to found the AAIF in December 2025.

The distinction between the two protocols is the point of putting them together: MCP standardizes how a single agent connects to outside tools and data sources, while A2A standardizes how separate agents — potentially built by different vendors — talk to each other and hand off work. A developer building a multi-agent system has, until now, had to treat them as separately governed projects even though most real systems need both.

The AAIF's own May 2026 membership update put its count at 190 organizations, with platinum members including Amazon, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft, and OpenAI — a rare list of frontier-lab rivals backing the same open-governance body. Whether A2A's move accelerates adoption beyond that platinum tier, or simply consolidates two standards that were already converging, is the open question the coming months will answer.

Neither Google nor the AAIF has published an exact transfer date for A2A's move from the Linux Foundation's general umbrella to the AAIF specifically, and Agent2Agent's existing specification, SDKs, and developer tooling are expected to carry over unchanged — this is a governance change, not a technical one. The practical test will be whether the two protocols start shipping joint reference implementations, rather than sitting under one foundation without integrating further.