Launch of CUVIC GPU Zero, an On-Premises Generative AI Platform Service

ITOCHU Techno-Solutions Corporation (headquartered in Minato-ku, Tokyo; Tatsushi Shingu, President and CEO; hereinafter CTC) has launched CUVIC GPU Zero, a generative AI platform service that enables the secure use of confidential data. The service provides a secure on-premises environment built on a closed network and is offered as a fully managed service that includes operations. Designed as a platform for safely utilizing generative AI while minimizing the risk of data leakage, the service features AI-dedicated hardware equipped with GPUs as well as capabilities for developing AI agents. Pricing starts at JPY 1.5 million as a front-end payment and JPY 750,000 (excluding tax) as a monthly fee. We will offer the service primarily to companies seeking to develop AI agents in-house, and aim to secure 30 orders over the next three years.

In recent years, as the use of generative AI has expanded across enterprises, the number of environments that allow companies to safely handle their own data—such as confidential information and business data—remains limited, driving growing demand for dedicated AI platforms. However, building an on-premises AI platform often requires significant investment, including the procurement of large numbers of GPU servers and peripheral equipment, as well as the costs and personnel needed for system construction, maintenance, and operation.

CUVIC GPU Zero is a managed service that combines AI-dedicated hardware equipped with NVIDIA GPUs and a platform featuring AI agent development capabilities, together with operational components such as a network environment, monitoring, incident response, and version upgrades. Under this service, CTC handles the equipment procurement, implementation, and ongoing maintenance and operations on behalf of customers. By building an on-premises environment, the service enables companies to securely use their confidential information and business data while minimizing the risk of data leakage. It supports simultaneous use by up to 50 users, making it suitable for deployment across multiple departments. The platform comes equipped with large language models (LLMs) and vision-language models*¹, allowing not only generating long-form texts and understanding of structured data, but also highly accurately taking in images such as charts and graphs. In addition, it incorporates Dify, a no-code AI agent development tool, allowing business departments with no programming expertise to rapidly develop AI agents tailored to their specific operational needs.

Leveraging our extensive track record in large-scale infrastructure development and experience in designing and utilizing AI models across a wide range of industries, CTC is responsible for GPU optimization, the design and construction of on-premises environments, platform monitoring and incident response, as well as the regular upgrading of software and AI models.

Going forward, the service will be further enhanced in line with the scale of use and expansion across departments, including the addition of GPU servers, network connectivity services, and new LLMs. CTC will also support the development of AI agents tailored to challenges in industry-specific tasks—such as production planning and equipment maintenance in the manufacturing sector, and transaction analysis and risk management in the financial sector—thereby providing an environment in which a wider range of companies can safely and confidently leverage generative AI in accordance with their own business needs.