www.industryemea.com
16
'26
Written on Modified on
Nokia Launches Industry-First Commercial AI-RAN Platform
Nokia has introduced a commercial AI-RAN platform built on software-defined anyRAN software and NVIDIA accelerated computing to optimize radio network spectral efficiency.
www.nokia.com
Nokia has announced the commercial release of its Artificial Intelligence Radio Access Network (AI-RAN) platform, establishing an architecture designed to host AI workloads directly within mobile network infrastructure. The software-defined platform allows telecommunication providers to increase uplink and downlink capacity from existing spectrum assets and radio infrastructure. Developed in collaboration with NVIDIA, the system mitigates reliance on traditional hardware upgrade cycles by implementing a cloud-native, programmable framework that scales operational throughput via targeted software delivery.
Spectral Efficiency Projections and Commercialization Timeline
The platform transitions mobile infrastructure from hardware-defined, fixed digital signal processing assemblies to an AI-native network architecture. Empirical network data indicates that the integrated AI-driven radio algorithms have delivered a baseline spectral efficiency improvement of more than 20%. Based on current product testing roadmaps, the architecture is projected to secure a 50% gain in total spectral efficiency by 2027, with targeted optimization models expanding capacity yields beyond 100% by 2028. This continuous optimization loop increases permissible network traffic volumes within dense macro cells while reducing the carrier's baseline operational cost per bit.
Pilot field deployments are scheduled to commence at the conclusion of this year, with full commercial system availability slated for 2027. The long-term product roadmap directly leverages NVIDIA’s programmable merchant silicon platforms. To decouple capability enhancements from physical asset upgrades, the platform introduces a commercial software subscription model. This structure allows telecommunication operators to systematically activate advanced optimization modules, threat mitigation features, and network scaling applications as they become available without incurring a hardware replacement premium.
Architectural Framework and Three Adoption Pathways
The common software architecture is powered by Nokia’s anyRAN software layered across NVIDIA’s accelerated computing engine. The software stack supports concurrent 4G, 5G, and subsequent network standards while maintaining strict alignment with open interfaces defined by Open RAN standards to preserve multi-vendor hardware interoperability.
To accommodate diverse deployment strategies, operators can adopt the AI-RAN framework via three distinct physical pathways:
- AirScale Capacity Plug-In Unit: Designed to safeguard pre-existing infrastructure investments, this path introduces a dedicated graphics processing unit (GPU) expansion card that plugs directly into active AirScale base station frames. The upgrade pathway integrates NVIDIA's computing hardware alongside AI-accelerated merchant silicon developed by Marvell to expand localized baseband capability.
- Standalone AI-RAN Node: A high-capacity, standalone hardware appliance engineered to host multi-generation workloads. The GPU-powered platform can be deployed as an independent network element, arranged within multi-node clustered configurations, or integrated side-by-side with conventional AirScale modules to form a single unified logical base station.
- Cloud-Native Server Solutions: Tailored for telecommunication networks embracing virtualized cloud topologies, this route utilizes GPU-powered Commercial Off-The-Shelf (COTS) server systems supplied through ecosystem hardware partners. The strategy unifies data center structural flexibility with the strict physical line-rate processing requirements demanded by AI-native basebands.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
