SoftBank And Red Hat's Power Optimisation for Data Centres
The global telecommunications industry faces mounting pressure to balance the growth of data demands against increasing energy costs and environmental concerns.
Now, as mobile networks evolve 5G and 6G technologies, the integration of AI with network infrastructure has emerged as both a solution and a challenge to these pressures – by offering performance improvements but increasing power consumption.
For instance, whilst AI is improving data processing, storage and security in data centres, the energy requirements of Graphics Processing Units (GPUs) – essential hardware for AI workloads – can be five to ten times higher than traditional computing hardware, creating new challenges for network operators.
However, tackling this problem, SoftBank, a major teco company and Red Hat, a provider of enterprise open source software solutions, have developed a system to monitor and optimise power consumption in data centres running virtualised Radio Access Networks (vRAN) and AI applications.
SoftBank's AI-RAN data centre expansion drives power management innovation
The solution integrates with SoftBank's "AITRAS" platform – a converged AI-RAN system and uses an orchestration tool to allocate computing resources to AI applications based on real-time power usage data – which achieves more efficient energy consumption.
- SoftBank and Red Hat developed AITRAS
- The solution leverages Kepler, an open-source project from Red Hat
- Service providers can now achieve greater sustainability and operational flexibility
AITRAS, which stands for AI and Telecom Radio Access System, represents SoftBank's vision for integrating AI capabilities with traditional mobile network infrastructure on a unified computing platform.
SoftBank is developing AITRAS as part of its strategy to deploy GPU server-equipped data centres throughout Japan – these facilities aiming to enhance mobile network performance through AI while supporting various AI applications.
However, operating these combined vRAN and AI applications requires substantial electrical power – which comes amid growing pressure to increase renewable energy adoption and reduce carbon emissions from fossil fuels.
This means that the concentration of data centres in Japanese metropolitan areas has also created challenges for power grid resilience, driving the need for more distributed energy load management.
These factors are what prompted SoftBank and Red Hat to create this power optimisation solution, leveraging their joint research and development partnership on AI-RAN technology that began in November 2024.
Red Hat OpenShift and Kepler project provide foundation for AITRAS power monitoring
The power monitoring capability derives from the Kepler project, which Red Hat has incorporated into its OpenShift container platform.
Kepler operates within Kubernetes environments – an open-source system for automating deployment and management of containerised applications.
The system collects power consumption metrics from servers in each data centre cluster and from individual applications. This energy usage data is then made available to the AITRAS Orchestrator software.
Kepler also delivers estimated power consumption information based on GPU utilisation, including resources partitioned using Multi-Instance GPU (MIG) technology, which allows a single physical GPU to be divided into smaller virtual GPU instances.
Therefore, SoftBank has enhanced its AITRAS Orchestrator by implementing Kepler into the virtualisation infrastructure built on Red Hat OpenShift – enabling the orchestrator to consider power consumption alongside traditional resource allocation factors such as data centre location, application priority and GPU allocation size.
The enhanced AITRAS Orchestrator can now additionally factor in real-time power consumption data when dynamically assigning AI applications to the most optimal resources.
For example, it can optimise resource allocation in ways including:
- In multi-cluster environments – it can monitor power consumption across clusters and assign applications to clusters with lower power usage
- Power consumption limits – can be set for each cluster, ensuring applications are deployed without exceeding the predefined threshold
- Renewable energy availability and carbon intensity metrics – can be set for each cluster, allowing applications to be deployed based on environmental impact considerations
The companies demonstrated this technology at Red Hat's exhibition stand during Mobile World Congress Barcelona 2025.
Ryuji Wakikawa, Vice President and Head of the Research Institute of Advanced Technology at SoftBank Corp, emphasises the critical infrastructure nature of both electricity and telco: “Electricity and telecommunications services continue to grow as critical infrastructure that supports society.
“By monitoring and predicting power consumption 'AITRAS' optimises equipment from an energy efficiency perspective while reducing risks through distributed deployment.
“The integration of telecommunications and power infrastructure paves the way for the future of AI-driven infrastructure.”
Meanwhile, Chris Wright, Chief Technology Officer and Senior Vice President of Global Engineering at Red Hat, highlights the collaborative effort: “Red Hat and SoftBank are committed to supporting the future of 5G and 6G use cases by bringing the combined power of AI and RAN to network orchestration and optimisation.
“With Red Hat OpenShift as a common platform, AI-RAN offers a pioneering approach to network operations for service providers to harness AI for improved resource efficiency and more sustainable power consumption, as well as supporting AI-enabled workloads across network environments.”
Explore the latest edition of Data Centre Magazine and be part of the conversation at our global conference series, Tech & AI LIVE and Data Centre LIVE.
Discover all our upcoming events and secure your tickets today.
Data Centre Magazine is a BizClik brand
