#Japan #Tokyo #GPUSOROBAN #NVIDIA HGX B200 #GPU Cloud Service

NVIDIA HGX B200: Revolutionizing AI Workloads in Japan

High-resolution, a leading provider of GPU cloud services, has officially begun pre-orders for the NVIDIA HGX B200, a groundbreaking platform designed for AI and machine learning applications. This announcement marks a significant milestone in the tech landscape, providing businesses and individuals with an affordable and powerful solution for their computing needs.

The NVIDIA HGX B200 leverages the latest Blackwell architecture, offering capabilities that can boost training speeds of Large Language Models (LLMs) by up to three times compared to its predecessor, the H100. This improvement is particularly beneficial for intensive tasks such as AI learning and deep learning, where computational demand is at its highest.

Affordable Pricing Plans

The company is providing competitive pricing plans for this advanced GPU instance: 240,000 yen per month per unit. This pricing strategy establishes High-resolution as one of the most cost-effective providers in the market, encouraging more users to access state-of-the-art AI tools and resources.

Specifications of the NVIDIA HGX B200

• - GPU/Node: NVIDIA B200 SXM x8
• - GPU Memory/Node: Total of 1440 GB
• - GPU Memory Bandwidth: Total of 62 TB/s (7.7 TB/s x8)
• - GPU Intercommunication (NVLink): Total of 14.4 TB/s (1.8 TB/s x8)
• - vCPU/Node: Equivalent to 112 cores
• - System Memory/Node: Equivalent to 4 TB
• - Storage/Node: 30 TB NVMe
• - Computational Performance (FP16 Tensor Core): Total of 36 PFLOPS (4.5 PetaFLOPS x8)
• - Computational Performance (FP8 Tensor Core): Total of 72 PetaFLOPS (9 PetaFLOPS x8)
• - Computational Performance (FP4 Tensor Core): Total of 144 PetaFLOPS (18 PetaFLOPS x8)
• - Region: Japan

Committed to AI Advancement

High-resolution aims to enhance accessibility to AI technologies for businesses and individuals by offering high-performance GPU platforms. The company's vision is to make Japan the country with the “most affordable AI computing power globally.” This strategic initiative aligns with broader goals to support the future of AI infrastructure in Japan.

About GPUSOROBAN

The GPUSOROBAN service accelerates extensive computational processes required for tasks like image generation AI and LLM. By reducing the costs of data center construction and operation, High-resolution is able to deliver NVIDIA’s high-performance GPU servers at lower costs. With over 2000 accumulated usage instances, GPUSOROBAN is widely adopted across various sectors, including IT, manufacturing, construction, and academic research.

Company Background

Founded in 2019, High-resolution operates one of Japan's largest GPU data centers located in Shika, Ishikawa. The company plans to open the first dedicated GPU data center for AI development in the Shikoku region by December 2024, recognized as part of the Ministry of Economy, Trade and Industry's Cloud Program Supply Assurance Plan. Additionally, in 2025, new GPU data centers will be established in Sagara, Saga, and Ayakawa, Kagawa, utilizing decommissioned schools for local revitalization and the development of generative AI.

Recognizing its outstanding contributions to the industry, High-resolution received NVIDIA’s “Best CSP Partner of the Year” award in June 2022 and was accredited by the Ministry of Economy, Trade and Industry in April 2024.

For more information about High-resolution and GPUSOROBAN, visit their corporate website at highreso.jp or access the GPUSOROBAN site at soroban.highreso.jp. For inquiries, email at soroban_support@highreso.co.jp.