Supermicro announced a partnership with Nokia to deliver an AI‑RAN solution that uses NVIDIA’s Aerial RAN computer platform. The joint offering places accelerated CPUs and GPUs in base stations and network hubs, letting carriers run AI workloads like speech and vision with minimal delay. Supermicro says the modular servers can be configured for 2U to 4U rack units and deliver up to 1.6 PFLOPS of AI performance. The first systems are shown at NVIDIA GTC Berlin and are expected to ship later this year, with broader availability in early 2027.
Why it matters
Network operators can host AI inference directly at the cell site, cutting response times for edge services such as real‑time video analytics.
