Huawei Open-Sources Telco AI Protocol

The A2A-T protocol was jointly released with global telecom partners at the TM Forum Accelerate Week on February 6, 2026. It is designated as IG1453 for its beta version and includes an enhanced prompt meta-model, IG1453A, aiming to create a unified framework for multi-agent collaboration in automated network operations. This initiative is a step towards highly Autonomous Networks, a key direction for the communications industry. Huawei's open-source project for A2A-T includes three core components: a Protocol SDK for standardized agent interaction, a Registry Center for agent authentication and management, and an Orchestration Center for low-code or no-code visual workflow creation. This move from a paper-based specification to providing practical developer tooling is intended to bridge the gap between industry consensus and real-world, multi-vendor deployment. The primary business problem A2A-T addresses is the inefficiency of integrating systems from different vendors. Proponents claim the standardized protocol can shorten system integration cycles from months to days, support complex cross-domain workflows, and lower interconnection barriers for a more sustainable collaborative ecosystem. This initiative enters a landscape of ongoing AI standardization efforts. The ITU-T established a Focus Group on Autonomous Networks (FG-AN) in late 2020 to lead pre-standardization studies. Concurrently, the ETSI Industry Specification Group on Experiential Networked Intelligence (ENI) has been defining a Cognitive Network Management architecture and has released specifications on use cases and requirements. Within the mobile network itself, 3GPP has been integrating AI/ML functionalities. Work in Release 18 focused on enhancements to data collection and signaling for network optimization, while Release 19 is studying AI/ML support for network slicing and Coverage & Capacity Optimization (CCO). This shows how intelligence is being embedded at various layers, from the radio access network to the overarching management plane where A2A-T operates. A2A-T is distinct from other agent communication standards like Google's A2A protocol, which was donated to the Linux Foundation. While Google's A2A targets the enterprise software layer for collaboration across business applications, Huawei's A2A-T is specifically designed for the telecom network layer, addressing the stringent reliability, security, and latency requirements of carrier-grade infrastructure. This open-source push is part of a larger strategic vision for an "Agentic Internet," where network traffic is increasingly dominated by AI agents communicating with each other, rather than just terminals connecting to a central service. This shift from traditional north-south traffic to more east-west agent interaction fundamentally reshapes network architecture requirements and lays the groundwork for 5G-Advanced and 6G.