SoftBank Unveils New AI Data Center Cloud Operating System

SoftBank has officially unveiled Infrinia AI Cloud OS, a cutting-edge software stack engineered to revolutionize the operation of AI data centers. This new platform aims to automate infrastructure management and deliver robust inference services for large language models. The introduction of this software signifies a strategic expansion for SoftBank, moving beyond traditional hardware offerings into the specialized domain of GPU cloud software.

The Infrinia AI Cloud OS is designed to manage a comprehensive range of tasks, spanning from basic BIOS configuration to advanced Kubernetes orchestration on powerful GPU platforms, including Nvidia’s state-of-the-art GB200 NVL72. This end-to-end automation is crucial for minimizing the complexities associated with running high-performance GPU cloud services, a significant pain point for many enterprises today. SoftBank asserts that this integrated approach will not only streamline operations but also substantially reduce the total cost of ownership and the inherent operational burden often linked with customized or in-house developed solutions.

According to Charlie Dai, VP and principal analyst at Forrester, this launch elevates SoftBank’s market position. Dai notes that the company is transitioning from a mere infrastructure operator to a formidable AI-native platform-level competitor. This shift reflects a broader industry trend towards more sophisticated, software-defined solutions in the rapidly expanding AI landscape. The platform’s ability to offer Kubernetes-as-a-Service (KaaS) in a multi-tenant environment, alongside Inference-as-a-Service (Inf-aaS) for large language models via APIs, positions it as a compelling offering for organizations seeking to enhance their GPU cloud capabilities.

Enhancing Enterprise AI Operations

The Infrinia AI Cloud OS provides two core services designed to tackle prevalent challenges in enterprise AI deployment. The Kubernetes-as-a-Service component offers extensive automation, covering everything from BIOS and RAID settings to operating systems, GPU drivers, networking, Kubernetes controllers, and storage. This comprehensive automation is vital for ensuring seamless and efficient management of complex GPU clusters.

Furthermore, the system intelligently reconfigures physical connectivity using Nvidia NVLink and optimizes memory allocation dynamically as users create, update, or delete clusters. This includes allocating nodes based on GPU proximity and NVLink domain configuration, a critical feature for minimizing latency and maximizing performance in demanding AI workloads. This sophisticated resource management ensures that computational power is utilized efficiently, a key factor in cost control and operational effectiveness for data centers.

Enterprises frequently grapple with the complexities of GPU cluster provisioning, Kubernetes lifecycle management, inference scaling, and infrastructure tuning, all of which demand deep technical expertise. Dai emphasizes that SoftBank’s automated approach directly addresses these challenges by simplifying BIOS-to-Kubernetes configurations, optimizing GPU interconnects, and abstracting inference into user-friendly API-based services. This strategic simplification allows development teams to concentrate more on model development and less on the intricacies of infrastructure maintenance, thereby accelerating innovation and deployment cycles within AI initiatives.

The Inference-as-a-Service component further empowers users by enabling the deployment of inference services with minimal effort. Users can select desired large language models without the need to configure Kubernetes or the underlying infrastructure. This service offers OpenAI-compatible APIs and boasts the capability to scale across multiple nodes, including high-performance platforms like the GB200 NVL72. To ensure security and reliability, the software incorporates tenant isolation through encrypted communications, automated system monitoring, failover mechanisms, and APIs for seamless integration with portal, customer management, and billing systems, providing a complete solution for modern AI inference needs.

Navigating a Competitive and Expanding Market

The introduction of Infrinia AI Cloud OS strategically positions SoftBank to capitalize on a rapidly expanding market, projected to grow significantly from $8.21 billion in 2025 to an estimated $26.62 billion by 2030. This substantial growth underscores the increasing demand for advanced AI infrastructure and specialized cloud services. However, SoftBank enters a highly competitive arena, facing established hyperscale cloud providers and specialized GPU vendors.

Major players like AWS, Microsoft Azure, and Google Cloud already offer robust managed Kubernetes services with extensive GPU support through their respective platforms—EKS, AKS, and GKE. Beyond these hyperscale giants, specialized providers such as CoreWeave, Lambda Labs, and RunPod have successfully built Kubernetes-native platforms that address similar operational challenges. These companies have carved out significant niches by focusing exclusively on high-performance GPU cloud solutions.

CoreWeave, for instance, operates a vast network of 45,000 GPUs and holds the distinction of being Nvidia’s first Elite-level cloud services provider. Lambda Labs also demonstrates strong market presence, reportedly generating $425 million in revenue in 2024 and offering competitive H100 instances. Against this backdrop, SoftBank’s software-centric strategy represents a pivotal shift, moving the competitive advantage from mere GPU availability to sophisticated platform automation.

Dai notes that as the demand for GPU-as-a-Service continues its rapid acceleration, differentiation increasingly hinges on intelligent orchestration, efficient inference abstraction, and integrated AI lifecycle tooling. The market is evolving beyond simply providing raw compute power toward offering comprehensive, full-stack AI-native cloud platforms. SoftBank’s entry with Infrinia AI Cloud OS reflects an understanding of this critical market evolution, aiming to provide a more holistic and automated solution for AI development and deployment. This approach seeks to provide a compelling alternative to existing offerings by prioritizing operational efficiency and advanced service integration.

Strategic Deployment and Future Outlook

SoftBank’s rollout strategy for Infrinia AI Cloud OS involves an initial deployment within its own GPU cloud services. This internal application will allow the company to thoroughly test and refine the software in a real-world, high-stakes environment before expanding its availability to external customers. Following this foundational phase, the Infrinia team plans to extend the software’s deployment to overseas data centers and various cloud environments, indicating a clear ambition for global reach and impact.

Junichi Miyakawa, SoftBank’s president and CEO, highlighted the foundational role of software in the advancement of AI infrastructure. He emphasized that physical components like GPU servers and storage must be complemented by sophisticated software that can integrate these resources, enabling them to be delivered flexibly and at scale. This perspective underscores SoftBank’s commitment to providing comprehensive solutions that go beyond hardware alone, focusing on the seamless integration and scalable delivery of AI capabilities.

The company reiterates that the Infrinia AI Cloud OS is specifically designed to reduce the total cost of ownership and alleviate the operational burden often associated with highly customized or in-house developed AI infrastructure solutions. While SoftBank has not yet disclosed specific pricing or availability details for external customers, the strategic direction is clear: to offer a highly automated, efficient, and scalable platform for the next generation of AI data centers. This initiative marks a significant step for SoftBank in solidifying its position within the burgeoning artificial intelligence ecosystem.