Overview
- T‑Mobile U.S. ran over‑the‑air tests using Nokia’s CUDA‑accelerated RAN on Nvidia platforms, concurrently handling AI applications and commercial 5G traffic.
- SoftBank reported a live trial reaching 16‑layer massive MIMO on a fully software‑defined setup, while Indonesia’s IOH advanced to pre‑commercial validation and demonstrated an AI‑powered 5G call with remote robot control.
- SynaXG executed fully software‑defined 4G/5G across sub‑6 GHz and millimeter‑wave on a single GH200 server, hitting 36 Gbps throughput with under 10 ms latency.
- Nvidia says 26 of 33 AI‑RAN Alliance demonstrations at MWC will run on its platforms, featuring resource‑sharing blueprints for safe GPU allocation and split‑inferencing across device, edge and cloud.
- A broad coalition and vendor ecosystem is aligning on open, software‑defined 6G foundations, with Nvidia open‑sourcing RAN libraries, joining the OCUDU initiative, and new integrations from QCT, Supermicro, WNC, Eridan and LITEON.