Details
- NVIDIA and Nokia announced major AI-RAN progress ahead of Mobile World Congress 2026 (March 2-5), with live field trials from T-Mobile U.S., SoftBank, and Indosat Ooredoo Hutchison demonstrating concurrent AI and RAN processing on GPU-accelerated platforms.
- Key milestones include T-Mobile's over-the-air demonstration using Nokia's CUDA-accelerated RAN software on NVIDIA Grace Hopper servers; SoftBank's industry-first 16-layer massive MIMO in software-defined 5G; and IOH's pre-commercial validation featuring Southeast Asia's first AI-powered 5G call with remote robotic control.
- SynaXG achieved the world's first AI-RAN implementation on millimeter wave (FR2) spectrum bands, delivering 36 Gbps throughput and sub-10ms latency on a single NVIDIA GH200 server running 4G, 5G, and agentic AI workloads simultaneously.
- MWC will showcase 26 of 33 AI-RAN Alliance demos built on NVIDIA AI Aerial, including innovations in AI-native air interfaces (DeepSig), split-inferencing for autonomous systems (SUTD), GPU resource orchestration (zTouch Networks), and dynamic algorithm switching (Northeastern University).
- A growing ecosystem of hardware partners—including Quanta Cloud Technology, Supermicro, WNC, Eridan, and LITEON—is delivering commercial off-the-shelf AI-RAN products supporting Nokia software and NVIDIA platforms, with NVIDIA open-sourcing Aerial CUDA libraries and joining the Linux Foundation's OCUDU Ecosystem.
Impact
AI-RAN is transitioning from experimental research to commercial-grade deployment, validating software-defined GPU acceleration as the foundation for next-generation wireless networks. Industry adoption by major operators across three continents signals accelerated 6G development timelines, with 77% of telecom leaders expecting faster deployment cycles than traditional standards trajectories. The convergence around NVIDIA platforms and open-source standards positions GPU-accelerated RAN as a competitive necessity for operators seeking to unlock new revenue streams through edge AI services and autonomous system support.
