Hitachi Vantara Expands 'Hitachi iQ' AI Infrastructure Solutions to Strengthen AI Sovereignty

Hitachi Vantara, a subsidiary of Hitachi, Ltd., announced today the launch of new enterprise models for its 'Hitachi iQ' AI infrastructure solutions in Japan. These models feature the latest NVIDIA Blackwell Ultra generation 'NVIDIA HGX B300' servers with liquid cooling. As a response to the growing importance of AI and data sovereignty, this on-premises solution significantly improves processing performance for large-scale AI models, enabling companies to accelerate AI development and deployment while maintaining control over their critical data.

With the rapid evolution of generative AI, there is an increasing demand for infrastructure that can process AI models faster and more efficiently. Furthermore, against the backdrop of strengthening economic security and rising cyberattack risks, the importance of 'sovereignty'—ensuring control over data and AI—has become paramount. Hitachi Vantara leverages its extensive experience, including building and operating internal generative AI infrastructure for the Hitachi Group, to provide optimal AI infrastructure solutions ranging from entry-level to enterprise models.

Key Features of the New Hitachi iQ Enterprise Models:

(1) High-Speed Processing for Large-Scale AI Models with Latest GPUs: The NVIDIA HGX B300, equipped with the latest NVIDIA Blackwell Ultra generation GPUs, supports high-speed processing for complex AI models such as Large Language Models (LLMs). GPU memory has been increased to a maximum of 2.1TB, 1.5 times that of the HGX B200. The 800Gb/s network accelerates communication between GPU clusters, significantly improving processing speed and stability.

(2) High-Efficiency Liquid Cooling: The liquid-cooled model of the NVIDIA HGX B300 reduces power consumption by up to 40% and rack space occupancy by up to 60% compared to air-cooled models. This design contributes to lower operational costs, higher installation efficiency in data centers, and stable performance over long periods by effectively managing GPU heat.

(3) High-Performance Distributed File Storage: 'Hitachi Content Software for File powered by WEKA (HCSF)' has achieved 'NVIDIA-Certified Storage' status at the NVIDIA Cloud Partner (NCP) level. This certification confirms its ability to meet rigorous requirements for extreme scalability (up to 10,000 GPUs), robust multi-tenancy, and strict Quality of Service (QoS) control. The integration of HCSF with NVIDIA AI Enterprise provides verified blueprints and comprehensive AI software lifecycle support.