- Samsung Electronics and KT Corporation have successfully demonstrated AI-based Radio Access Network optimisation on a live commercial network, showing potential advances for 6G development.
- The validation highlights technical gains but also raises questions about real-world operational complexity, costs, and the path to broad commercial adoption.
What happened: AI-driven network optimisation validated on a live network
Samsung Electronics and KT Corporation have successfully demonstrated AI-based Radio Access Network (AI RAN) optimisation on KT’s commercial mobile network in Seongnam, Gyeonggi Province. The validation, announced this week, marks the first time the technology has been tested under real operating conditions rather than controlled simulations.
The field trial covered around 18,000 users across areas with varying signal-quality challenges. AI RAN adjusts radio parameters at an individual-user level, rather than applying uniform optimisation to an entire cell. By learning from usage patterns and movement behaviour, the system predicts connection instability and proactively adjusts network configurations to reduce failures.
According to the companies, users who previously experienced repeated connection problems saw a significant drop in failures after the AI RAN-driven adjustments were applied. Service quality also improved for other users within the same geographic zones, indicating broader cell-level benefits.
Samsung’s research division and KT’s Future Network Laboratory plan to expand testing to additional commercial environments, viewing AI-driven optimisation as a foundational component of future 6G architectures.
Also Read: AWS launches Project Rainier with 500,000 Trainium2 chips
Also Read: Telus and Samsung push Open RAN upgrade
Why it’s important
The validation is a meaningful milestone for next-generation RAN development, particularly as global vendors and operators explore how artificial intelligence can respond to the rising complexity of future networks. AI RAN is expected to play a major role in 6G, where billions of connected devices, new mobility patterns and ultra-low latency requirements will demand real-time, automated decision-making within the network.
However, several strategic questions remain. Scaling AI-based optimisation across large, multi-vendor networks introduces operational and cost challenges. The technology requires extensive data processing infrastructure, raising concerns around efficiency, interoperability and long-term maintainability.
Another uncertainty is commercial viability. Operators around the world have been cautious about AI-driven network features that improve internal performance but do not create new revenue streams. Until AI RAN enables capabilities that customers or enterprises are willing to pay for, wide-scale monetisation remains uncertain.
The results achieved in a controlled South Korean environment may also be difficult to translate directly to markets with more diverse geographies, hardware ecosystems or regulatory conditions. The breakthrough is technically promising, but significant work remains before AI RAN becomes a standard component of mainstream networks.
