Google is moving governance of Agent2Agent (A2A), the open protocol it created for AI agents built by different vendors to talk to one another, from its general home at the Linux Foundation into the Agentic AI Foundation (AAIF) — the Linux Foundation's directed fund focused specifically on agentic AI infrastructure. Axios reported the move on August 17, and it brings A2A under the same governance umbrella as Anthropic's Model Context Protocol (MCP), the other dominant agent standard.

Two protocols, one home

MCP and A2A solve different problems. MCP, which Anthropic open-sourced in November 2024, standardizes how an individual agent connects to external tools and data sources. A2A, which Google introduced in April 2025, governs how separate agents — potentially built on different models and by different companies — discover each other's capabilities and exchange tasks. Anthropic donated MCP to the Linux Foundation and helped establish the AAIF in December 2025 alongside OpenAI and Block, with Google, Microsoft, AWS, Cloudflare and Bloomberg joining as supporters. Google had kept A2A under the Linux Foundation's broader portfolio rather than folding it directly into the new foundation at that time; this transfer closes that gap.

The Linux Foundation says A2A has been adopted by more than 150 organizations in its first year, with production deployments across major cloud platforms and support from companies including Microsoft, AWS, Salesforce, SAP and ServiceNow. The AAIF itself has grown quickly since its December launch, expanding its base of enterprise supporters as more companies commit to building on open agent standards rather than proprietary, vendor-specific integration layers.

Why it matters

For developers building multi-agent systems, having MCP and A2A share one governance structure reduces the risk of the two standards drifting apart or duplicating effort as agentic AI scales — a common worry in fast-moving open-source ecosystems with overlapping corporate backers. It also signals that the industry's biggest labs, which compete fiercely on model capability, see a shared interest in agreeing on how their agents interoperate at the protocol level, similar to how competing companies historically cooperated on networking or web standards even while competing on products. For enterprises evaluating multi-vendor agent deployments, consolidated governance is a modest but concrete signal that A2A and MCP are being treated as durable infrastructure rather than single-company projects that could be abandoned or changed unilaterally.