SynaXG Achieves Breakthrough In AI-Native 5G Radio Networks

KUALA LUMPUR, March 2 (Bernama) -- SynaXG announced major technical breakthroughs in artificial intelligence (AI)-native radio access networks, demonstrating concurrent operation of 5G FR1, 5G FR2 and AI workloads on NVIDIA AI-RAN platform with real-time, policy-driven GPU orchestration.

According to SynaXG in a statement, the results confirm that software-defined, AI-native RAN can deliver carrier-grade performance and continuous round-the-clock operation, ready for commercial deployment.

Using a single NVIDIA GH200 platform, SynaXG achieved carrier-grade 5G FR1 performance, simultaneously operating 20 x 100 Megahertz (MHz) 5G NR cells.

bootstrap slideshow

The system delivered over 36 gigabits per second (Gbps) of aggregated throughput with sub-10 millisecond latency, supporting up to 1,200 connected user equipment per cell, validating that AI workloads can run concurrently without compromising network Key Performance Indicators (KPIs).

SynaXG also achieved the world’s first carrier-grade FR2 virtualised RAN implementation running alongside FR1 and AI workloads on a shared GPU platform. The system delivered high throughput with end-to-end latency as low as five milliseconds, meeting requirements for dense urban and enterprise deployments.

Its real-time orchestration software dynamically allocates Multi-Instance GPU partitions based on live RAN and AI performance metrics. GPU resources are continuously rebalanced across FR1 for coverage, FR2 for capacity, and AI workloads without service interruption, ensuring deterministic performance while maximising infrastructure utilisation.

bootstrap slideshow

The AI-RAN system operated continuously under sustained load, validating the stability, scalability, and resilience required for carrier environments. SynaXG confirmed that NVIDIA GPU-accelerated AI-RAN maintains predictable performance while adapting dynamically to changing traffic and workload demands.

Built on NVIDIA AI Aerial platforms, SynaXG’s AI-RAN software-defined architecture supports deterministic network performance alongside AI processing on shared GPU infrastructure. The company plans to extend the platform to additional NVIDIA platforms, including DGX Spark, highlighting its “write once, run anywhere” portability.

SynaXG will showcase these AI-RAN breakthroughs as part of the AI-RAN Alliance Working Group 2 demonstrations at the AI-RAN Alliance booth and at the SynaXG booth at Mobile World Congress 2026 in Barcelona.

-- BERNAMA