In a move that signals a paradigm shift for global telecommunications, Nokia and SoftBank Corp. have announced a strategic agreement to launch a commercial field trial of Artificial Intelligence-Radio Access Network (AI-RAN) technology in Japan. Scheduled for 2027, the trial represents a pivotal moment in the transition from traditional, hardware-centric network architectures to software-defined, AI-native ecosystems. By integrating Nokia’s anyRAN software with SoftBank’s proprietary AITRAS platform—powered by NVIDIA’s accelerated computing—the companies aim to validate the feasibility of running radio and AI workloads on a unified, programmable infrastructure under real-world, commercial traffic conditions.
Main Facts: The Architecture of Tomorrow
The upcoming trial is designed to stress-test the convergence of connectivity and compute. At its core, the project utilizes Nokia’s AI-native anyRAN Cloud RAN software, which will be deployed on SoftBank’s AITRAS platform. The AITRAS framework serves as the core orchestration engine for the AI-RAN ecosystem, ensuring that radio resources and artificial intelligence processes operate in synergy.
Crucially, the hardware foundation for this trial is provided by the NVIDIA Aerial RAN Computer (ARC) platform. This high-performance setup incorporates the NVIDIA Grace CPU, NVIDIA RTX PRO 4500 GPU, and NVIDIA ConnectX-8 networking interfaces. This configuration provides the high-bandwidth, low-latency connectivity necessary to handle the intensive computational demands of modern RAN workloads.
By leveraging the NVIDIA CUDA ecosystem, the project moves beyond a standard network upgrade. It establishes a software-defined foundation that allows for continuous innovation, effectively creating a sandbox for the evolution toward AI-native 6G networks. The integration will see Nokia’s MantaRay network management solution working in tandem with the AITRAS Orchestrator to ensure seamless resource management across the entire stack.
Chronology: From Lab Bench to Live Network
The road to the 2027 field trial is the culmination of years of iterative development in telecommunications infrastructure.
- Pre-2024 (Foundational Research): Both Nokia and SoftBank invested heavily in virtualization and cloud-native RAN (Radio Access Network) research. During this phase, the focus was on decoupling software from proprietary hardware, a prerequisite for the AI-RAN model.
- 2024 (Collaborative Alignment): Nokia and SoftBank formalized their collaborative roadmap, focusing on how to integrate NVIDIA’s accelerated computing nodes into existing carrier-grade infrastructure.
- 2025–2026 (Integration & Laboratory Validation): The current phase involves rigorous laboratory testing, where Nokia’s anyRAN software is tuned to operate within the specific constraints of the AITRAS platform. Engineers are currently working on optimizing the orchestration layer to ensure that AI applications do not interfere with standard carrier-grade connectivity requirements.
- 2027 (Commercial Field Trial): The scheduled launch of the field trial will see the technology moved into a live commercial environment in Japan. This will be the first instance where AI and RAN workloads are co-located and managed on the same platform under heavy, unpredictable, real-world consumer traffic.
- Post-2027 (Path to 6G): The data gathered from the 2027 trial will serve as the blueprint for the commercial rollout of AI-RAN, acting as the bridge toward 6G, where AI will be fundamentally embedded into the air interface of mobile networks.
Supporting Data: Why AI-RAN Matters
As the demand for generative, agentic, and physical AI applications skyrockets, the traditional limitations of cellular networks have become a bottleneck. Currently, RAN infrastructure is primarily designed for data throughput and mobility. However, the rise of the "intelligent edge"—where AI inference must happen close to the user to minimize latency—requires a different approach.
According to industry metrics, the integration of AI and RAN on a common platform offers several distinct technical advantages:
- Infrastructure Utilization: By sharing the compute resources of the NVIDIA ARC platform between RAN processing and AI tasks, operators can significantly increase the "return on compute." When network traffic is low, the platform can pivot its capacity toward AI model inference or data processing.
- Latency Reduction: By placing AI workloads at the edge (within the base station infrastructure), the time required for data to travel to a centralized cloud server and back is virtually eliminated. This is critical for applications like autonomous robotics and real-time immersive digital experiences.
- Spectral Efficiency: AI-driven beamforming and resource allocation can optimize the air interface in real-time, allowing networks to handle higher user densities without compromising quality of service.
The hardware specifications provided by the NVIDIA ecosystem are a testament to the scale of this requirement. The NVIDIA Grace CPU and RTX PRO 4500 GPU are purpose-built to handle the massive parallelism required for AI, while simultaneously maintaining the deterministic, sub-millisecond response times required by 5G standards.
Official Responses: A Shared Vision
The collaboration between these industry titans reflects a unified belief in the necessity of network transformation.
Pallavi Mahajan, Chief Technology and AI Officer at Nokia, emphasized the transformative nature of the agreement:
"Our agreement with SoftBank marks an important step toward the future of AI-native networks. By bringing AI and RAN workloads onto common, programmable infrastructure, AI-RAN has the potential to unlock new levels of spectral efficiency while providing the distributed intelligence needed to support physical AI. As the boundaries between network and compute continue to blur, this distributed infrastructure can become a powerful platform for both connectivity and AI."
Ryuji Wakikawa, Senior Vice President and CTO at SoftBank Corp., highlighted the practical milestones being achieved:
"Our planned field trial with Nokia will mark a significant milestone in AI-RAN’s transition from concept to execution. We will integrate Nokia’s AI-native anyRAN into our AITRAS platform as its RAN component… We are excited to begin this next phase of validation with Nokia, paving the way for a new era of networks that deliver both connectivity and AI services."
Implications: The Future of the Networked World
The implications of the Nokia-SoftBank trial extend far beyond the borders of Japan. The industry is currently witnessing a fundamental shift in how telecommunications operators view their assets. Historically, an operator’s greatest asset was its spectrum and tower coverage. In the era of AI-RAN, the operator’s greatest asset is its distributed compute footprint.
The Rise of Physical AI
The trial aims to support use cases that are currently theoretical or in their infancy. Physical AI—the integration of artificial intelligence into robots, drones, and autonomous vehicles—requires a network that is not just "fast," but "intelligent." These devices must interact with their environment in real-time, often processing massive datasets on the fly. The AI-RAN architecture provides the necessary "brain" at the edge to enable this, effectively turning the network into a distributed supercomputer.
The Evolution Toward 6G
The industry is already looking past 5G. While 5G was designed for enhanced mobile broadband and IoT, 6G is being architected as an AI-native standard. By testing these technologies today, Nokia and SoftBank are effectively prototyping the core features of 6G. The successful orchestration of AI and RAN on the AITRAS platform will prove that the network can handle heterogeneous workloads, a capability that will be mandatory for the complex, hyper-connected landscape of the 2030s.
Economic and Operational Shifts
For operators, the ability to support new revenue streams—such as providing edge-computing-as-a-service to enterprises—is vital. If a factory requires low-latency AI for its robotics, the mobile operator can now provide that service directly through the existing RAN infrastructure. This creates a new, high-value service tier that moves operators up the value chain, shifting them from "dumb pipe" providers to essential partners in the AI economy.
Conclusion
The 2027 field trial is more than a technical test; it is a declaration of intent. By choosing to co-develop these solutions, Nokia and SoftBank are signaling that the future of telecommunications is not in hardware, but in the intelligent, programmable, and efficient orchestration of data. As the trial nears, the entire global telecom sector will be watching, as the results will likely dictate the standard for how the world connects, computes, and creates in the age of artificial intelligence.
