In earlier generations, artificial intelligence (AI) functioned largely as an optimisation layer added to existing systems. Emerging network visions now describe AI as an embedded architectural component. The growing visibility of AI across sectors is not driven by models alone, but by the evolution of underlying networks, cloud systems and data centres that are beginning to function as an interconnected fabric. Without this underlying network foundation, even the most advanced AI applications would struggle to operate at scale or deliver meaningful real-time outcomes.
From structured automation to generative intelligence
Traditional AI systems have already been widely used in telecom environments, particularly for structured and well-defined tasks. These systems rely on statistical and supervised learning models and continue to perform effectively in scenarios requiring reliability and deterministic outputs. The rapid growth of AI-driven applications is already reshaping network traffic dynamics. Multimodal assistance and interactive digital services are increasing traffic volumes while introducing more bursty, unpredictable and uplink-intensive usage patterns. This represents a structural shift rather than an incremental increase in demand.
Generative AI (GenAI), however, expands this operational scope by moving beyond analysis to synthesis. Instead of only classifying or predicting based on existing data, it generates new outputs by learning patterns from large data sets. This allows networks to evolve from reactive systems into context-aware and adaptive environments.
In operational settings, GenAI can interpret multi-source logs, generate summaries, simulate alternative scenarios and recommend remediation steps. In customer service, it enables conversational systems that handle open-ended queries and maintain contextual continuity, replacing rigid script-based interactions. It also supports the creation of synthetic data sets and digital twins of network environments, enabling more advanced simulation and planning capabilities.
Within radio access networks, GenAI can enhance channel prediction, generate synthetic radio data sets and support digital twin modelling of radio environments, reducing reliance on extensive physical testing. In core networks, it enables automated policy generation, intelligent slicing and advanced traffic modelling. This signals a broader shift from rule-based self-organising systems to predictive, context-aware and self-learning network frameworks supported by distributed inference across devices and edge nodes. Further, the relationship between GenAI and telecom networks is becoming deeply interdependent, particularly as the industry advances towards 5G Advanced and 6G.
GenAI is also expanding the scope of simulation-driven network planning and optimisation. Digital twins of network environments allow operators to model demand scenarios, simulate network behaviour and test performance under varying operational conditions. Synthetic data generation further supports preparedness for rare or complex scenarios, reducing dependence on extensive real-world testing while enabling faster innovation cycles.
In optimisation contexts, GenAI is increasingly explored as a solution generator for complex network problems. Traditional mathematical models used for environment estimation, resource allocation and monitoring are reaching their limits as network architectures grow more heterogeneous and dynamic. Emerging approaches, including diffusion-based models and sequential reinforcement methods, offer adaptive optimisation capabilities.
Furthermore, the move towards autonomous networks is accelerating as AI-assisted operations become integral to forecasting, fault management and orchestration. With greater data availability and improved telemetry from modern network systems, AI agents are already being deployed to automate routine operational processes. This is particularly important given the coexistence of multiple generations of network technologies, expanding spectrum usage and increasing operational complexity.
Compute and connectivity
Two fundamental forces are driving this transformation: connectivity and compute. Connectivity continues to expand rapidly across successive generations of mobile technology, fibre expansion and satellite integration, while compute capabilities have scaled through increasingly powerful processing architectures capable of synthesising massive volumes of data. Data serves as the shared currency between these forces, with networks transporting information and compute systems processing it into actionable insights.
When connectivity and compute converge, the impact becomes multiplicative. Compute enhances network awareness, enabling real-time optimisation, predictive decision-making and personalised service delivery. At the same time, connectivity allows intelligence to flow across devices, vehicles, edge environments and user applications, bringing computation closer to the point of use. This creates a distributed intelligence model in which inference, transmission and action occur seamlessly and in real time across digital ecosystems.
Deploying GenAI models at the edge enables real-time immersive applications, autonomous systems and intelligent industrial use cases that cannot rely solely on centralised computing. Collaborative inference architectures, where computational workloads are shared between devices, edge nodes and core infrastructure, help manage latency while optimising resource utilisation.
This distributed model reflects a broader shift in computing, where intelligence flows across devices, vehicles, enterprise systems and digital platforms. Connectivity plays a critical role in enabling this flow, ensuring that inference, transmission and execution occur seamlessly in real time. As a result, intelligence becomes pervasive rather than centralised, embedded within both physical and digital environments.
Challenges
As AI becomes embedded within network architecture, trust and resilience emerge as foundational priorities. Risks such as data poisoning, adversarial manipulation, privacy leakage and model vulnerabilities carry significant implications when integrated into telecom systems that support essential services.
Trust is closely tied to cybersecurity, fraud prevention and ethical data usage, and it ultimately determines user acceptance of AI-enabled services. Without strong trust foundations, technological advancements alone cannot ensure widespread adoption. Responsible deployment also requires robust data governance systems, including anonymisation standards, privacy safeguards and transparent data-sharing mechanisms.
Further, excessive reliance on fully automated decision-making can reduce transparency in operational processes, especially in mission-critical environments where service continuity is essential. Maintaining human oversight and manual override mechanisms remains crucial to ensure accountability, reliability and strategic control over critical infrastructure.
Therefore, the rapid expansion of AI has highlighted the need for balanced policy focus across the digital ecosystem. While considerable attention has been directed towards compute infrastructure, cloud expansion and data centres, the network layer that carries AI intelligence requires equivalent strategic emphasis.
Supportive policy frameworks that encourage infrastructure investment, streamline regulatory processes and foster innovation are essential to sustaining this ecosystem. Equally important is collaboration across industry, academia, start-ups and policymakers to ensure harmonised development and avoid fragmentation. Standardisation and interoperability, long central to telecom evolution, remain critical as AI frameworks and network architectures continue to converge.
Future outlook
India’s digital ecosystem provides a distinctive context for the convergence of GenAI and future networks. With large-scale connectivity, a vast broadband user base and a rapidly expanding digital economy, the country offers a unique environment for scaling AI-driven solutions. Existing digital infrastructure has already demonstrated how technology aligned with public purpose can enable widespread adoption and accessibility. As intelligence is layered on to this infrastructure, the focus shifts towards inclusive and context-aware deployment that reaches diverse populations and geographies.
As GenAI adoption accelerates, networks are witnessing a structural shift in how traffic is generated and consumed. This transformation places new demands on network infrastructure, particularly in terms of latency, uplink capacity and real-time responsiveness.
However, to address these changes, networks must evolve towards adaptive architectures that can sense, act and adjust in real time. AI-assisted operations in traffic forecasting, network management and optimisation are becoming integral to handling increasing complexity.
Future networks, from 5G-Advanced to 6G, must therefore be designed not only for higher capacity but for intelligence and automation. Earlier connectivity models focused on connecting people and devices, whereas the next phase increasingly centres on connecting intelligence. This requires networks capable of handling dynamic, bursty and unpredictable traffic patterns while maintaining reliability and performance across diverse applications.
A future-ready digital economy depends on coordinated development across compute, storage, cloud and connectivity, as these elements operate as an integrated system. At the same time, as telecom infrastructure assumes the role of critical national infrastructure, the balance between automation and human oversight remains essential to ensure accountability, transparency and operational resilience in increasingly autonomous network environments.
Based on a discussion among Shyam Prabhakar Mardikar, President and Group Chief Technology Officer (Mobility), Reliance Jio Infocomm Limited; Syed Tausif Abbas, Senior Deputy Director General, DoT; Gurvinder Singh Ahluwalia, Founder and Chief Executive Officer, Digital Twin Labs, USA; Jeanette Whyte, Head of Public Policy (APAC), GSMA; Manoj Gurnani, Chief Technology Officer and Head of Strategy, India, Nokia; and Col P K Choudhary, Program Director, HCL, at the India AI Impact Summit 2026.
