Why AI is altering planning for 6G mobile networks

AI is altering the planning of future 6G mobile networks as operators grapple with uncertain capacity demands.

The Next Generation Mobile Networks Alliance published a study detailing how operator networks must adapt to support AI applications. The publication consolidates perspectives from mobile network operators to guide ongoing standardisation studies within the 3rd Generation Partnership Project.

The publication makes it clear there’s a necessity for networks to evolve beyond basic connectivity to handle new service demands. The alliance advocates an evolutionary approach that builds upon existing fifth-generation architectures to ensure long-term investment protection and interoperability.

Current mobile data consumption consists mostly of video applications, which account for 70-75 percent of total traffic. AI interactions are primarily text-based and have a modest impact on mobile network traffic today.

However, multi-modal AI applications could alter traffic patterns substantially, especially if consumers adopt augmented reality glasses or enterprises deploy autonomous vehicles. These devices require continuous uploads of environmental images and sensor data, reversing the traditional downlink-heavy traffic pattern.

Operators must design flexible networks to accommodate this uncertainty. The standardisation process should explore mechanisms for adjusting the ratio of uplink to downlink traffic without requiring major standard revisions. Base stations may need enhanced uplink slot occurrences to maximise transmission opportunities.

“As 6G standardisation enters a critical phase, the rapid growth of AI and AI agents presents both opportunities and challenges for mobile network operators,” said Laurent Leboucher, Chairman of the NGMN Alliance Board and Orange Group CTO and EVP Networks.

“Given the variety of future AI use cases and applications, it is essential that 6G standards enable adaptability without forcing disruptive architectural changes. Flexibility will be critical to accommodate evolving AI use cases across devices, networks, and regions.”

To monetise these infrastructure investments, telecom companies are exploring new charging models. Token-based charging is one proposed method to account for specific resource consumption. Tokens would correspond to defined units of bandwidth or edge computing capacity. This mechanism promotes fair cost allocation and incentivises efficient resource usage among users and network operators.

Enterprise applications require dynamic networking to support collaboration among physical AI agents. Operators will need to support short-lived, mission-specific private networks on-demand. Example use cases include collaborative industrial robots and autonomous vehicle fleets. These network groups must adapt to changing service requirements and environmental conditions while minimising manual provisioning overhead.

Edge computing resources are necessary to support low-latency inference. User equipment has limited computing power, and centralised cloud processing introduces latency bottlenecks.

Distributed computing allows real-time processing and efficient resource utilisation near the data source. Information sharing across these different domains requires a unified data framework. Without this framework, data remains siloed, preventing AI applications from sharing models and intermediate results across heterogeneous devices.

How 6G mobile networks must evolve to support AI

The transition to 6G mobile networks will build upon the existing 5G service-based architecture. Network functions must evolve to interact with third-party software components and AI agents. The Model Context Protocol (MCP) is one potential mechanism for integrating agents with network function resources.

Multi-vendor interoperability remains a primary concern for operators. The standardisation process must determine whether communication protocols for agents should be formally standardised or adopted through de facto industry practices. Ensuring interoperability minimises integration complexity for mobile operators.

Integrating AI into the radio access network requires selective implementation. Large data volumes at the medium access control layer can benefit from algorithmic processing. Functions that already operate close to optimal limits, such as channel coding or basic synchronisation, are not expected to see large performance gains. Executing inference locally at the edge is necessary to meet strict latency requirements and avoid negatively impacting radio access network functions.

“The proliferation of AI use cases, particularly those with autonomous, task-driven capabilities, is rapidly reshaping how networks are built and operated,” said Anita Döhler, CEO of the NGMN Alliance.

When operators evaluate the benefits of introducing new capabilities to enhance an existing service, they should assess the net carbon dioxide impact alongside the net financial impact.

Testing models in real network environments is essential, as algorithms trained on idealised datasets may exhibit limitations across different deployment conditions. Networks must maintain support for non-AI alternatives to ensure reliability and openness.

The core network domain requires application programming interface evolution to manage third-party software requests efficiently. Legacy systems often lack the hardware capacity to interpret new control requests. Operators must design converged management interfaces to control both legacy and new systems during hardware replacement cycles.

Trust frameworks are required for agent-to-agent communication to identify malicious content and block unauthorised actions. Compliance and lawful interception capabilities must remain intact to meet regional regulatory obligations. Implementing strong encryption and integrity checks protects sensitive prompts and personal customer data.

By aligning early on standardisation priorities, operators can ensure future 6G mobile networks are sustainable and value-driven.

See also: Google plans $15B subsea cable to boost AI infrastructure

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the IoT Tech Expo and Cyber Security & Cloud Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.