Every major evolution of core network architecture has driven explosive growth in new connections and services. From 2G/3G IP architecture enabling mobile voice, to 4G CUPS triggering data growth, and 5G cloud-native architecture powering gigabit speeds—network transformation has always catalyzed industry progress. Therefore, the Next-Gen Core Network must adapt to the unique connectivity demands of the AI era.

Redefining Connectivity: From "Human and Things" to "Massive AI Agents"

Over the next decade, mobile communications will fully enter the AI era. With AI Agent capabilities doubling every four months, rapidly penetrating devices, operational scenarios, and business workflows. Huawei's Intelligent World 2035 report that global AI Agents will reach 900 billion by 2035—roughly 100 times the human population.

Mobile network connections will fundamentally shift toward massive AI Agents. Delivering task-level SLAs and massive uplink capacity for real-time multimodal inference will be essential for future networks.

Agentic Services: The Next Typical Scenario for Mobile Networks

As AI-enabled devices and applications achieve widespread adoption, Agentic services are emerging as typical application scenarios for mobile networks. Consequently, the intrinsic value of mobile networks will shift from simply transferring raw data (Bits) to delivering task-oriented outcomes (Tokens), primarily driving three key scenarios:

1. AoNR (Agent over New Radio) for Task Latency Assurance: Keeps single-task Token transmission latency under 50ms, guaranteeing end-to-end task latency below 400ms for cloud-offloaded AI inference.
2. WCN (Wide-area Computing Network) for Computing Efficiency Gains: Interconnects terminal (e.g., autonomous vehicles), edge, and in-network computing resources to deliver elastic, on-demand AI inference.
3. ACN (Agent Computing Network) for Collaboration Security Assurance: Complex, dynamic problems require more than a single agent. ACN enables intelligent agent collaboration across domains and ecosystems.

Defining Next-Gen Core Network via AI Native Architecture

Traditional signaling-driven networks cannot interpret a terminal's true task intent. Building an intent-driven "terminal-network collaborative mechanism" will serve as a foundational capability for next-gen core network.

In June 2025, 3GPP launched a dedicated research project on next-generation architecture across 24 key study items. Led by Huawei and industry consensus, the industry is aligning around an AI Native architecture defined by three core features:

• Interfaces (Signaling → Intent): Directly communicates personalized terminal needs (experience, compute, topology) for easy service invocation.
• Functions (Two Planes → Three Planes): Introduces a Data and AI Plane (DAP) to interpret task intent and dynamically match network resources.
• Workflows (AI Overlay → AI Native): DAP collaborates with the Control and User Planes via Agent-Based Interfaces (ABIs), embedding AI decision-making directly into foundational workflows.

Shaping the Future Together

Simply overlaying AI onto legacy networks cannot unleash the full potential of the mobile ecosystem spanning silicon, terminals, OS, infrastructure, and applications. Huawei calls on global industry partners to jointly advance the standardization of AI Native architecture—centered on Intent Interfaces and DAP—to build a robust digital foundation for the era of massive AI Agents.