6G AI Architecture Reshaping Next-Generation Network Design

The development of 6G is bringing artificial intelligence closer to the fundamental design of mobile networks. While earlier generations have increasingly used AI and machine learning to optimise specific functions, the emerging 6G approach is considering intelligence as something that can be embedded across the network architecture itself. This represents a shift from adding AI capabilities to networks toward designing networks that are inherently capable of supporting intelligent processes.

This is making 6G AI architecture an important area of 6G development. The International Telecommunication Union has included “artificial intelligence and communication” among the six usage scenarios of its IMT-2030 framework, while ubiquitous intelligence is identified as one of the framework’s overarching principles. These developments indicate that AI is being considered alongside core network capabilities rather than only as an external application.

AI is Moving Deeper into Network Design

The distinction between AI-assisted and AI-native networks is becoming increasingly important. An AI-assisted network can use machine learning to optimise a particular process, such as resource allocation or traffic management. An AI-native network goes further by incorporating intelligence into the architecture, allowing AI-related capabilities to influence how network functions communicate, operate and adapt.

The standards process is beginning to reflect this shift. 3GPP Release 20 includes dedicated studies examining AI protocols for 6G, alongside separate work on 6G core-network control-plane and user-plane protocols. This indicates that AI is being considered at multiple architectural layers rather than being confined to radio optimisation.

For 6G AI architecture, this creates a different design challenge from simply deploying more machine-learning models. Future networks will need to determine how models interact with network functions, how data is exchanged, how inference is managed and how intelligent capabilities can be exposed to applications and services.

The Architecture is Being Developed Around Intelligence

The move toward an AI-native model also changes the relationship between data and network functions. AI systems depend on continuous access to data for training, inference and adaptation, meaning future network architectures may need to treat data availability and processing as fundamental capabilities.

ITU’s work on AI-native networks is examining this broader architectural question, including terminology, use cases and potential network architectures for embedding AI more deeply into telecommunications systems. At the same time, 3GPP’s emerging 6G studies are examining specific protocol and core-network requirements that could support more intelligent network behaviour.

This creates a progression from the AI-enabled optimisation explored in current 5G-Advanced systems toward a more deeply integrated model. 6G AI architecture is therefore not simply about adding AI to more parts of the network. It is about changing how network intelligence, data and communication functions are organised.

Key Takeaway: AI is expanding from individual network optimisation functions into multiple architectural layers as 6G standardisation begins to define how intelligent capabilities can be embedded into future networks.

The significance of this transition extends beyond individual AI applications. As standards work moves into protocols, core-network functions and architectural design, 6G AI architecture is increasingly becoming part of the technical foundation being developed for the next generation of mobile networks.

AI is Expanding Across the 6G Network Architecture

The architectural role of artificial intelligence is becoming broader as 6G standardisation moves beyond individual radio functions and into the core network and management layers. This development is creating a network environment in which AI could influence how functions communicate, how resources are managed and how services interact with network capabilities.

This makes 6G AI architecture increasingly dependent on the integration of intelligence across multiple network layers. 3GPP Release 20 work includes studies on AI protocols as well as separate investigations into 6G core-network control-plane and user-plane protocols. Together, these areas indicate that AI is being considered as part of the mechanisms through which future networks operate rather than simply as an application running on top of them.

AI is Expanding Beyond the Radio Access Network

The move toward an intelligent core is particularly significant. In conventional network architectures, AI can analyse information generated by network functions and provide recommendations or optimisation outputs. A more deeply integrated architecture could allow intelligent capabilities to interact directly with network functions and influence decisions across the control and user planes.

This could affect functions such as traffic management, service orchestration, resource allocation and network exposure. Instead of treating each process as an isolated optimisation problem, the architecture could allow information and intelligence to move between functions based on changing network conditions and service requirements.

The emergence of dedicated AI protocol studies is important in this context. Protocols determine how network functions exchange information and coordinate activity, so introducing AI-related mechanisms at this level could provide a more structured way of integrating intelligent capabilities throughout the network.

6G AI architecture will therefore require more than high-performing models. It will also require defined mechanisms for deploying, managing and communicating with those models within the network.

Network Intelligence is Becoming More Distributed

Another architectural consideration is where intelligence should reside. Future networks could distribute AI capabilities across devices, radio access infrastructure, edge computing resources, core networks and centralised cloud environments. This can allow processing to take place closer to where data is generated while reducing unnecessary movement of information across the network.

Distributed intelligence also creates new requirements around model coordination, data access and computing resources. AI models may need to operate with different levels of autonomy while still maintaining consistent policies and service objectives across the wider network.

3GPP’s broader Release 20 work on 6G management and orchestration is relevant to this development because intelligent functions will need to be coordinated across a network that is increasingly distributed and dynamic. AI therefore becomes both a capability used by the network and a factor shaping how network resources are organised.

The architecture will also need to accommodate different AI workloads without allowing intelligence-related processing to undermine network efficiency or reliability. This makes the relationship between communication, computing and data an increasingly important architectural question.

AI is Moving Toward the Core of Network Operations

The direction of standardisation suggests that future 6G systems are being designed with intelligence in mind at several interconnected layers. 6G AI architecture could bring together AI protocols, core-network functions, orchestration and distributed computing into a more coordinated operating model.

The implications extend to network service exposure as well. Intelligent capabilities may need to be made accessible to applications while maintaining appropriate control over data, performance and security. This could create a closer relationship between network intelligence and the services that depend on it.

The emerging model is therefore different from simply adding machine learning to existing functions. AI is increasingly being considered in relation to the protocols, interfaces and management mechanisms that determine how the network itself operates.

As these studies progress, 6G AI architecture is likely to depend on how effectively intelligence can be integrated without creating fragmented AI systems. The objective is a network in which intelligence can operate across functions while remaining coordinated, manageable and responsive to changing requirements.

AI-Native Architecture Could Define the Next Network Operating Model

The development of 6G is moving AI beyond individual optimisation functions toward a more integrated role in network architecture. Emerging standards work is examining AI protocols, core-network functions, management and orchestration, creating the foundations for networks in which intelligence can operate across multiple layers.

This makes 6G AI architecture increasingly important to the design of future mobile systems. The challenge will be to coordinate AI capabilities across distributed infrastructure while maintaining performance, security, reliability and efficient use of network resources.

As 6G standardisation progresses, 6G AI architecture could reshape how networks process data, manage resources and adapt to changing service requirements. The transition therefore represents more than an expansion of AI use. It points toward a network model in which intelligence becomes an integral part of how the system is designed and operated.

6G Development Bringing AI into the Core of Network Architecture

AI is Moving Deeper into Network Design

The Architecture is Being Developed Around Intelligence

About us

Subscribe us

System

Resources

Archive Links

Subscribe