NTTPC, Getworks and Fixstars begin offering liquid-cooled GPU container data center package in Japan
Редакция ЦОД:Главное
NTTPC Communications, Getworks and Fixstars have begun offering the liquid-cooled GPU container DC-PKG in Japan, an integrated package for deploying AI infrastructure on customer-provided sites and power systems. The offering combines a containerized data center, liquid-cooled NVIDIA GPU servers, monitoring and operational support.
- The three companies have begun offering an integrated liquid-cooled GPU container data center package in Japan.
- The package is designed to use customer-provided land and electrical equipment.
- The infrastructure includes a 40-foot container, CDU-chiller, ceiling and in-row cooling, security, networking, monitoring and operational support.
- A small configuration with one GPU server has reference pricing from JPY300 million upfront and JPY750,000 per month, excluding tax.
- The announcement concerns product availability; it does not confirm a specific customer deployment or operational data center.
NTTPC Communications, Getworks and Fixstars have launched the Liquid-Cooled GPU Container DC-PKG, a packaged AI-infrastructure offering designed for deployment using a customer’s land and electrical equipment.
The service combines a 40-foot containerized data center with liquid-cooled NVIDIA accelerated-computing servers, security and networking equipment, integrated monitoring and operational support. The container infrastructure includes a CDU-chiller, ceiling-mounted and in-row air-conditioning equipment, cameras and other supporting facilities.
Available server configurations can include HPE ProLiant Compute XD685 systems with NVIDIA HGX B300. The offering is intended as an integrated deployment model rather than a standalone server product, combining physical infrastructure, cooling, compute and operations in a single package.
Reference pricing for a small configuration with one GPU server starts at JPY300 million in initial costs and JPY750,000 per month, excluding tax. These are starting prices, with individual pricing determined separately.
For operators and enterprises deploying local AI environments, the model could simplify coordination across container infrastructure, liquid cooling, GPU servers and monitoring suppliers. Its use of customer-provided land and power infrastructure may also be relevant where local data processing is required. The announcement describes the start of product availability, not the commissioning of a specific data center or a deployment for a named customer.
The package addresses cooling and integration challenges associated with high-density GPU infrastructure by bringing container, liquid-cooling, server, monitoring and operational elements into one offering. This could reduce procurement and deployment complexity for enterprises seeking on-premises AI capacity or local processing of sensitive data.
- Impress Watch — クラウド&データセンター / 9 сентября 2026 г. · ЯпонскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.