Software-centric architectures, unified data strategies and practical AI use cases are turning operator ambition into operational reality
Artificial intelligence (AI) has become the defining technology narrative across the telecommunications industry. For mobile network operators (MNOs), the challenge, and opportunity, is to apply AI in ways that deliver measurable operational gains today while laying the foundation for more autonomous networks in the future.
Samsung Networks sees many of the most promising AI applications in telecoms emerging where data, software and operational pragmatism intersect—leading the way to network autonomy.
Starting with solvable problems
While Samsung is working to actualize fully autonomous networks by 2027, the company’s near-term research has focused on high-impact use cases, including enabling more energy efficient operation of the radio access network (RAN). Power optimization has proven to be fertile ground for AI because it combines clear economic incentives with abundant, structured network data.
Samsung found that AI-driven optimization could significantly outperform existing manual approaches already used by operators. Samsung engineers were able to validate a maximum power consumption savings of 35% for a key operator, by using individual cell traffic patterns to adjust the time at which elements of the network turned off.
By adapting network behavior to local usage conditions, AI enables more granular and aggressive power-saving strategies without compromising performance. These kinds of targeted wins help operators build confidence in AI systems and create momentum toward broader automation.
Applying AI to complex radio decisions
Beyond energy savings, Samsung is using AI for complex RAN optimization challenges that traditionally rely on deep human expertise. One example involves balancing capacity and coverage through antenna parameter optimization, which led to around a 9% uplift in throughput without any loss of connectivity.
These results, initially validated in simulation and now progressing toward real-world deployment, demonstrate how AI can augment even highly-specialized engineering domains. However, as AI becomes more pervasive across the network, new challenges may emerge around coordination, prioritization and conflict resolution between multiple automated systems.
Data and software as foundational enablers
For operators seeking to move beyond pilots, success with AI depends on two foundational pillars: data and software. Volume and quality of data is critical, especially when working toward end-to-end optimization, and the key is to aggregate data from all parts of the networks into one, central place.
Samsung’s approach is to break down traditional data silos across RAN, core and management domains; this data is then unified in a common layer where AI models can be applied holistically. Softwarization is the other side of the coin and unlocks a new level of flexibility. By decoupling software from hardware across the network, operators gain the flexibility to dynamically place workloads where they deliver the best performance and efficiency.
From automation to autonomy
Samsung’s CognitiV Network Operations Suite is designed to support this incremental journey, allowing operators to start with narrowly focused optimizations and expand toward more autonomous behavior over time. The emphasis is on trust and gradual transformation rather than abrupt disruption.
Over time, as AI systems move from offering recommendations to taking direct action, operators can reduce operational complexity, lower operating costs and reallocate human expertise toward higher-value tasks. In Samsung’s view, that progression from assisted automation to trusted autonomy is where AI’s long-term value in telecoms will ultimately be realized.
Table of Contents
