Google's Agent2Agent protocol, the company's open standard for letting independent AI agents communicate with one another, formally joined the Agentic AI Foundation on Aug. 17, placing it under the same neutral governance as Anthropic's Model Context Protocol. The move brings the two most widely adopted agent standards — one governing how agents talk to tools and data, the other how agents talk to each other — under a single roof for the first time.
The Agentic AI Foundation, hosted by the Linux Foundation, was formed in December 2025 around three founding contributions: MCP from Anthropic, the goose agent framework from Block, and OpenAI's AGENTS.md specification. It has grown quickly since, expanding from roughly 49 founding members to more than 250 today, according to the foundation, with Platinum-tier backers including Amazon Web Services, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft and OpenAI — direct competitors that rarely agree to share governance of shared infrastructure.
Two protocols, two jobs
MCP and A2A solve different halves of the same problem. MCP, which Anthropic donated to the Linux Foundation when the foundation launched, standardizes how an individual AI application reaches out to external tools, files and data sources. A2A, first released by Google in April 2025 and backed since launch by more than 150 partner organizations including AWS, Cisco, Microsoft, Salesforce, SAP and ServiceNow, instead governs the horizontal link between two separate agents — how they discover each other, negotiate a task and exchange credentials across organizational boundaries.
A2A enables agents to easily discover, delegate, and collaborate across frameworks and siloed platforms, without custom integrations.
Rao Surapaneni, VP and General Manager, Business Application Platforms, Google Cloud
The foundation's technical leadership described the addition in similar terms. "A2A represents an important step toward an open, interoperable future for AI agents," said Manik Surtani, the Agentic AI Foundation's chief technology officer, in the announcement.
Why the merger of governance matters
A2A already has production mileage: Google cites deployments inside Huawei's Celia assistant on HarmonyOS, Tencent's WeChat, Google Cloud, Microsoft's Azure AI Foundry and Amazon's Bedrock AgentCore, plus a joint effort with PayPal on agent-driven payments. Moving it into the same governance structure as MCP means both specifications now go through the same technical committee, the same security review process, and the same neutral trademark and IP framework — reducing the odds that either protocol quietly forks into incompatible, vendor-controlled versions as adoption scales.
For developers building multi-agent systems, the practical upside is fewer decisions about which company's rules to trust. Enterprise engineering teams have spent much of the past year wiring together separate, sometimes competing, integration layers for tool access and agent-to-agent messaging; shared governance doesn't make those two jobs the same, but it does mean a single foundation is now responsible for keeping their interfaces from drifting apart as both specifications keep evolving.
