At MWC Barcelona 2026, Singapore-based company SynaXG demonstrated the ability to simultaneously handle 5G FR1, 5G FR2 bands, and AI workloads on a shared NVIDIA GPU platform, showcasing how operators can deploy radio access networks. The demo utilized NVIDIA's AI Aerial platform and a single GH200 system, sharing GPU resources between RAN and AI tasks.
SynaXG ran 20 100 MHz 5G NR cells, achieving an aggregated throughput exceeding 36 Gbps and latency below 10 milliseconds, with each cell supporting up to 1,200 connected devices. The company also showcased the first carrier-grade FR2 virtualized RAN, delivering end-to-end latency as low as 5 milliseconds, meeting the requirements for dense urban and enterprise deployments.
A key part of the demonstration was SynaXG's real-time orchestration software, which dynamically allocates GPU partitions, rebalancing resources between FR1 coverage, FR2 capacity, and AI workloads without service interruption. This improves computational resource utilization while maintaining stable RAN performance.
SynaXG stated that the AI-RAN system operates 24x7 under continuous load, proving its stability and resilience. The platform adapts to changing traffic and AI demands, maintaining predictable performance. The company is extending the platform to other NVIDIA systems, including DGX Spark, and developing AI-for-RAN capabilities covering optimization, automation, and energy efficiency.
Huang Xin, CEO of SynaXG, said: "With recent industry-leading breakthroughs, SynaXG has demonstrated that AI-RAN can deliver carrier-grade FR1 and FR2 performance on shared NVIDIA AI infrastructure, achieving continuous 24x7 operation." Soma Velayutham, Vice President of AI and Telecom at NVIDIA, stated: "Software-defined architecture is key for next-generation wireless networks. SynaXG's benchmark 5G performance on the NVIDIA AI-RAN platform proves that operators and enterprises can achieve the flexibility and agility of cloud-native computing while maintaining the carrier-grade throughput and performance-per-watt necessary for commercial 5G services."
SynaXG will showcase the setup as part of the AI-RAN Alliance Working Group 2 demonstration during the event, indicating that AI-RAN technology is moving from concept to deployment planning.
