1
Agentic AI is transferring from analysis to implementation, with rising trade consensus that future cellular networks should turn out to be AI-native. The query now’s whether or not operators will embrace the architectural modifications required to assist that transition.
For the third yr operating, the AI Core for Agent Communication Community (ACN) Seminar at MWC26 Shanghai confirmed the potential of clever community brokers to maneuver past principle and towards sensible implementation. AI is being integrated into telco operations, however it’s important that the telco neighborhood doesn’t fall behind the enterprise and hyperscaler sector.
If that is to be the case, agentic AI has architectural necessities for the 6G core, fairly than as a software program overlay on legacy community features. Operators are producing positive aspects from integrating AI into their assurance, planning, buyer care, and operations. Nevertheless, there may nicely be a two-tier system between telcos that use AI to optimize present domains and those who redesign their core community round AI-native rules.
AI-native is a strategic requirement
Telcos and the third Technology Partnership Mission (3GPP) requirements physique are gearing up for 6G. The 6G core is predicted to evolve from a comms-centric 5G ‘service-based structure’ (SBA) towards a extra cognitive and intent-driven structure. The target will not be solely effectivity and simplification, but in addition a ‘community for AI’ mannequin the place AI supervises and coordinates visitors administration, service, and useful resource optimization.
A number of telcos are already taking proactive steps to combine AI, generative AI, and even agentic AI precursors into their operations. Over the previous three years, NTT Docomo’s 6G core priorities have emphasised sustainability, effectivity, buyer expertise, ‘community for AI’, and ubiquitous connectivity. Orange has already been scaling AI in community environments, with 150 use instances rolled out in 2024 that reportedly generated €200 million in worth, whereas its CogNet work with Nokia Bell Labs and IBM goals to maneuver community administration from reactive to proactive cognitive operations.
KPN Fieldlabs has additionally been exploring a holistic SBA plus AI ‘service administration and orchestration’ (SMO) framework, together with an AI core occasion with each agent-based and service-based interfaces. These operators are demonstrating how AI can reshape not solely community administration, however even the core structure itself.
How agentic AI matches into the 6G core
Agentic AI has the potential to be a strong enabler for telcos, however the ROI may very well be muted if AI is an overlay or constrained to area of interest, end-user purposes. Agentic AI brokers should be capable of interpret targets, discover sources, set up belief, coordinate computing, and full duties throughout a variety of domains: finish customers, purposes, units, and networks. Having attended the previous three occasions, it’s hanging to see the extent of maturity. 5G networks have been constructed primarily for transmitting information, whereas agentic 6G situations are about executing and finishing duties.
This agentic AI imaginative and prescient is standards-based. The position, sources, and coverage entry for AI brokers are already being mentioned throughout a number of 3GPP SA2 6G architectural key points, together with ‘non-access stratum’ (NAS) – the core control-plane signaling layer between a tool and the community core, location companies, ‘AI for community’ (AI4NET), distant and close to area sensing, and the info framework.
Agentic AI will not be a distinct segment characteristic; it’s designed to deal with the visitors and intents from a number of domains (see determine 1).
Determine 1 | 6G logical framework overview
Intent interfaces
In typical mobile packet core structure, NAS signaling is very structured and rule-based. Units request predefined companies, periods, or mobility procedures. Within the AI-native structure, the NAS evolves towards ‘intent-based’ communication. A tool or agent now not solely requests a inflexible quality-of-service (QoS) profile; as an alternative, it could possibly sign the target it’s attempting to realize, permitting the 6G AI core to find out how finest to meet the request.
This creates a extra direct hyperlink between consumer aims, service logic, and community habits. As a substitute of reacting to visitors flows after the actual fact, the community can interpret intent and proactively coordinate connectivity, compute, safety, and latency.
In the course of the seminar, Turkcell highlighted its R&D in intent-based comms. One 6G state of affairs introduces devoted new 6G core community features to deal with and fulfill intent-based requests, in addition to investigating the affect of control-plane AI brokers on NAS identifiers and signaling flows. Moreover, Turkcell has centralized service operations heart information that depend on its Teknocan LLM, unified dashboards throughout spine, entry, core, and ICT domains.
ACN is a flagship take a look at case
On the seminar, China Cellular, Huawei, ZTE, CAICT, and Vivo highlighted the event work achieved, thus far, on ACN. The ACN interfaces with id, belief, routing, discovery, session administration, functionality publicity, compute collaboration, and low-latency communications. The ACN framework teams this into three areas: agent connection companies, agent empowerment companies, and agent collaboration companies.
China Cellular’s briefing described how its analysis institute has been in a position to join a number of agent varieties, together with robotics, drones, AR glasses, and digital brokers. Its Net4Agent work centered on trusted entry, dynamic subnetworks, computing offload, and multi-agent coordination, whereas Agent4Net addressed agentic AI-for-network features and AI agent protocol design.
The imaginative and prescient for ACN’s intent-based communication is additional demonstrated by its internet-of-agents open community protocol (AONP) and its implementation throughout a pilot in Zhejiang in China. The pilot was in a position to deal with features comparable to registration, identifier allocation, authentication, authorization, and routing. There may be potential so as to add AI brokers to deal with digital id, agent discovery, job session administration, task-level service-level agreements (SLAs), and AI-compute coordination.
Conclusion
Determine 2 | AI software program funding and 6G core cap-ex outlook
The speed of improvement in AI is transferring at an exponential tempo. ABI Analysis’s personal analysis sees AI utilization within the wider financial system as accelerating. The quantity of funding in AI sensing (audio/imaginative and prescient), generative AI, predictive AI, and pure language processing (NLP) is predicted to develop from$174 billion in 2025 to $467 billion by 2030.
There may be important danger that telcos may very well be left behind, managing the bit pipe however not securing client-facing, enterprise-orientated alternatives. Forecasting the anticipated funding into the 6G core community is difficult given among the binary choices going through the telco neighborhood. If there may be sustained dedication from the telco neighborhood for a extra proactive strategy to AI within the core upgrades, funding within the 6G core community ought to develop from$5 billion in 2029 to $24.5 billion by 2034.
Thus far, a major proportion of telcos have been conservative, and even perhaps overly cautious, in regards to the adoption of the most recent innovation tech. Nevertheless, with AI, they’re doubtlessly going through one other paradigm shift within the technological panorama that may very well be as impactful as the online 2.0 period that introduced in WhatsApp and iMessage.
The rise of WhatsApp and social media severely dented the SMS income streams and telcos’ aspirations. Telcos that embrace AI-native core evolution, intent-based interfaces, and agent communication frameworks will place their networks on the coronary heart of the 6G service mannequin and be capable of meet their clients’ AI expectations – each customers and enterprises, and their respective agentic AI brokers.
