Huawei Outlines AI-Native Core Network Plan for the Age of AI Agents

Why Huawei says the core network has to change

Mobile networks have traditionally supported voice calls, text messaging and mobile internet access. According to Huawei, the emerging agentic era requires support for a massive number of AI agents, which in turn requires a core network architecture that can understand intent, make intelligent decisions and ensure the end-to-end completion of AI-driven tasks.

Huawei believes an AI-native core network will be essential to this shift. By embedding AI into interfaces, functions and procedures, operators could move from providing connectivity alone to delivering intelligent network capabilities for the next generation of digital services.

The company points to its own Intelligent World 2035 report, which says the number of AI agents globally could reach 900 billion by 2035, approximately 100 times the world's population. Huawei argues that the core network must be able to support the long-term growth of new connections and services. As AI terminals and applications spread, agentic services will become typical mobile network scenarios, driving the network from transporting data bits to enabling task-oriented tokens, according to Huawei.

Three agentic service scenarios

Huawei identified three typical scenarios for assuring the agentic experience:

• Agent Over New Radio (AoNR) for task latency assurance: Portable AI devices face limits in computing power, battery capacity, size and cost, which makes end-to-end latency essential. AoNR is designed to keep end-to-end task latency below demanding thresholds.
• Wide-Area Computing Network (WCN) for computing efficiency gains: A WCN can interconnect computing resources through mobile networks, enabling task-based computing scheduling and standardized computing services.
• Agent Computing Network (ACN) for collaboration security assurance: Huawei says most complex tasks will involve multiple AI agents working across domains, enterprises and ecosystems rather than a single agent, which is constrained by model capabilities, context limitations, role conflicts or sequential processing challenges.

For the ACN, Huawei lists three capabilities for cross-domain and cross-ecosystem collaboration: cross-domain trusted digital identity, skill-based agent discovery and routing, and task-driven dynamic networking.

Interfaces, functions and procedures

To bridge terminals and networks, Huawei says the next-generation core network architecture should have three key characteristics.

• Interfaces, from signaling to intent: Intent-based interfaces let terminals pass personalized task requirements, including networking, computing and experience needs, directly to the network.
• Functions, from C/U planes to C/U+DAP planes: A dedicated Data & AI Plane (DAP) would sit alongside the traditional control and user planes as an intelligence layer. It would interpret task intents and dynamically align resources in real time.
• Procedures, from AI overlay to AI-native: AI decision-making would be embedded deeply in fundamental network procedures to coordinate connectivity and computing seamlessly.

Two stages of evolution and what it means for operators

Huawei outlined two stages for the AI-driven core network. In stage one, the network improves customer experiences by overlaying AI capabilities. Stage two is the AI-native next-generation core network. Huawei says this enables "enhanced connectivity", where task-oriented AI services ensure task completion, and "beyond connectivity", where the network provides capabilities beyond communication, such as computing services and facilitating collaboration between agents.

The company says adapting to these services will require restructuring the network architecture, particularly interaction mechanisms, data architecture and computing architecture. The transition also needs collaboration across the ICT ecosystem, including terminals, chips and networks.

For operators, Huawei's message is that terminal-network synergy and an advanced core network architecture could enable new connections and services and open up new business opportunities. The release is a statement of Huawei's vision and does not give deployment timelines, products or commercial terms.