Overview
- T‑Mobile and Ericsson ran Ericsson Cloud RAN software on NVIDIA AI infrastructure in Bellevue, demonstrating portable RAN software that gives operators more compute choices.
- SynaXG reported an FR2 millimeter‑wave AI‑RAN trial using a single NVIDIA GH200 server with 20 component carriers, reaching 36 Gbps throughput and under 10 ms latency.
- Ericsson and Intel announced a collaboration spanning RAN, core and edge to accelerate ecosystem readiness for AI‑native 6G.
- SoftBank detailed its Telco AI Cloud that combines a GPU data‑center platform, an AI‑RAN MEC layer and the Infrinia AI Cloud OS, and it open‑sourced the AITRAS orchestrator’s Dynamic Scoring Framework.
- NVIDIA expanded its open ecosystem by open‑sourcing Aerial RAN libraries and joining the Linux Foundation’s OCUDU effort, while vendors including Samsung, Nokia and IOH showcased AI integrated with vRAN in demos that validate feasibility short of commercial rollout.