Huawei is outlining an AI-native core network architecture designed to support the growing use of AI agents across mobile networks. The proposed approach moves beyond conventional connectivity by combining intent-based interfaces, integrated data and AI processing, and intelligent resource coordination. Huawei also identifies three supporting concepts—Agent over New Radio (AoNR), Wide-Area Computing Network (WCN) and Agent Computing Network (ACN)—to address task latency, computing efficiency and secure collaboration between AI agents.
Three Network Capabilities Target Agentic Services
Huawei identifies three service scenarios for next-generation mobile core networks, each addressing a different constraint associated with AI agents.
Agent over New Radio (AoNR) focuses on task-latency assurance. Portable AI devices can face limitations in processing power, battery capacity, size and cost. Huawei’s proposed approach aims to coordinate network capabilities to help keep end-to-end task completion within demanding latency thresholds. The company has not specified universal latency targets or provided benchmark results in the supplied announcement.
Wide-Area Computing Network (WCN) focuses on computing efficiency. The concept connects computing resources through mobile networks, enabling task-based scheduling and standardized computing services. This could allow AI workloads to use resources beyond a device’s own processing capacity, depending on available infrastructure and the demands of each task.
Agent Computing Network (ACN) addresses collaboration between AI agents operating across different domains, enterprises and ecosystems. Complex tasks may require multiple specialized agents to discover one another, exchange information and coordinate actions securely. Huawei identifies three enabling capabilities: trusted digital identities across domains, skill-based agent discovery and routing, and task-driven dynamic networking.
Together, these concepts position the network as more than a transport layer. They envision connectivity, computing and agent coordination operating as integrated services, although their implementation will require further technical development and interoperability work.
From Connectivity to Intent-Based Network Architecture
Huawei proposes three architectural changes to support this transition.
First, network interfaces would move from conventional signaling toward intent-based interactions. Instead of communicating only connection parameters, a device could express a task’s requirements, including connectivity, computing resources and expected user experience.
Second, the architecture would introduce a dedicated Data and AI Plane (DAP) alongside the traditional control and user planes. This proposed layer would interpret task intent and coordinate resources dynamically across network and computing infrastructure.
Third, AI would become embedded in fundamental network procedures rather than operating as an overlay. Huawei argues that deeper integration would help coordinate connectivity and computing throughout the task lifecycle.
These proposals align with a wider industry discussion about AI-native networks and the evolution of 6G. The International Telecommunication Union has established a focus group to examine architectural requirements and opportunities for AI-native telecommunications networks.
The distinction between a conceptual architecture and a standardized, commercially deployed network remains important. Operators will need interoperable interfaces, reliable performance, security controls and clear mechanisms for managing competing tasks before such capabilities can be deployed at scale.
A Two-Stage Transition Toward AI-Native Networks
Huawei describes the evolution in two stages. The first adds AI capabilities to existing network architectures to improve customer experiences and network operations. The second restructures the core network around AI-native capabilities.
In this second stage, Huawei envisions two broad service categories: “enhanced connectivity,” where networks help ensure AI-driven tasks meet their requirements, and “beyond connectivity,” where operators provide computing services and facilitate collaboration between agents.
The commercial opportunity for telecom operators could extend beyond traditional data plans to differentiated connectivity, computing-resource access and agent-oriented services. However, monetization will depend on customer demand, service-level agreements, operational costs and the ability to coordinate resources across network and cloud providers.
Huawei’s forecast of 900 billion AI agents by 2035 comes from its own Intelligent World 2035 outlook, rather than an independently established industry consensus.
The central challenge is therefore not simply connecting more devices. It is designing networks that can interpret tasks, allocate resources and coordinate AI services reliably. Huawei’s proposal outlines one possible path, but its eventual impact will depend on standards development, ecosystem cooperation and real-world deployment evidence.
Market Landscape
Telecommunications networks are evolving as AI workloads introduce new requirements for latency, computing capacity, data exchange and service coordination. The industry is exploring how network intelligence can move beyond traffic optimization toward intent-aware services and AI-assisted orchestration.
Huawei’s Intelligent World 2035 outlook projects 900 billion AI agents by 2035. This is a company forecast, not an independently confirmed prediction.
The broader direction is also reflected in industry research. The ITU’s AI-Native Networks focus group is examining the architectural changes needed to embed AI into telecommunications systems. Ericsson has likewise discussed AI-native services and agent collaboration as potential elements of future 6G core networks.
Huawei’s proposed AoNR, WCN and ACN concepts address task latency, computing-resource coordination and secure agent collaboration. Turning these concepts into commercial services will require standards alignment, interoperability, security safeguards and measurable service-level guarantees.
Top Insights
- Huawei proposes an AI-native core network that interprets task intent and coordinates connectivity, computing resources and AI-agent collaboration.
- AoNR, WCN and ACN target three challenges: task latency, computing efficiency and secure collaboration across agent ecosystems.
- A proposed Data and AI Plane would complement traditional control and user planes, enabling more dynamic task-aware resource coordination.
- Huawei forecasts 900 billion AI agents by 2035, highlighting the company’s view of future network demand rather than an established industry consensus.
- Telecom operators could develop computing and agent-coordination services beyond connectivity, but commercial adoption will depend on standards, security and demonstrated customer demand.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
