SoftBank Corp. and Nokia to Enable Execution
of External AI Workloads on AI-RAN

Background

SoftBank is developing the AITRAS Orchestrator to gain real-time visibility into demand for computing resources used for AI processing and RAN (Radio Access Network) control, and to dynamically and flexibly allocate those resources within AITRAS.

Until now, SoftBank has used the AITRAS Orchestrator to run internal AI workloads within the company alongside RAN control on the same virtualization platform, which dynamically optimized the allocation of computing resources. However, when the usage is limited to RAN control and internal AI workloads, fluctuations in demand due to differing time periods or traffic conditions can result in underutilized computing resources. To leverage AITRAS as infrastructure with higher added value, a mechanism was required that could accommodate not only internal but also external workload demand and flexibly allocate computing resources. In particular, as the demand for AI processing continues to grow, it can be expected that new revenue opportunities and capital investment efficiency will improve with telecommunications operators offering distributed computing resources as AI processing services for external customers.

About this Initiative

In this initiative, SoftBank has expanded the functionality of the AITRAS Orchestrator by integrating Nokia Bell Labs's AI platform, Nokia AI-RAN External Compute Engine, making it possible to optimize the allocation of computing resources across both internal and external AI workloads. This enables AI-RAN infrastructure to be utilized as a unified AI execution platform, regardless of whether workloads are internal or external. External customers can access computing resources on the AI-RAN infrastructure on demand without their own computing platforms. For telecommunications operators, this creates a new business model centered on providing computing resources for AI processing by leveraging existing communications infrastructure.

The AI-RAN platform will continue to evolve into infrastructure that can flexibly allocate computing resources in response to external AI demands and deliver enhanced value to customers. As AI demand accelerates, expanding usage beyond traditional internal AI and RAN applications will enable further growth in the form of new revenue opportunities.

This initiative realizes the Execution of External AI Workloads, one of the use cases defined in the AI-RAN Platform & Infrastructure Orchestrator by the AI-and-RAN Working Group of the AI-RAN Alliance. In addition, by presenting a concrete architecture for resource utilization that includes external AI demand, this effort is expected to promote further discussions on AI-RAN architectures and use cases in the foreseeable future.

Looking Ahead

A demonstration of this technology is scheduled to be presented at Nokia's booth at MWC Barcelona 2026. Building on the newly added capabilities of the AITRAS Orchestrator, SoftBank will continue to promote initiatives toward the practical deployment of AI-RAN, while expanding support for diverse external AI use cases and conducting validation in large-scale environments.

Ryuji Wakikawa, Vice President and Head of the Research Institute of Advanced Technology at SoftBank Corp., said:
"In AI-RAN, it is important to maximize the value of compute resources. We enhanced the AITRAS Orchestrator to allocate resources to external AI workloads, allowing us to leverage those resources as a new revenue source."

Pallavi Mahajan, Chief Technology and AI Officer at Nokia commented:
"This collaboration with SoftBank marks an important milestone in advancing AI-RAN from concept to real-world application. As demand for AI processing accelerates globally, this project showcases how distributed network resources can be harnessed to deliver scalable, efficient, and sustainable AI services. We are proud to work with SoftBank to drive innovation that unlocks new business models for operators and expands the value of AI RAN for the entire industry."