GPUSOROBAN

Try It Free for 30 Days

Experience High-Performance GPU Computing at Unbeatable Value
—at less than half the cost of AWS.

Experience cutting-edge AI compute power without
building your own servers or data centers. Start instantly from anywhere.

Try it free

FEATURES

Value Proposition

Plug-and-Play
AI Supercomputing

No setup. No infrastructure. Get immediate access to NVIDIA H200 GPUs in the cloud.

Enterprise-Grade
Performance
at Unmatched Cost

Train LLMs, fine-tune generative AI, or run complex simulations — all at a fraction of AWS pricing.

Built for Regions
Without
Cloud Infrastructure

Our secure Japan-based servers enable companies in underserved regions to leapfrog into high-performance AI development.

USE CASE

AI and Machine Learning Engineers

Manufacturing, Universities, and Research Institutions

AI Startups’ Development Teams

Companies Developing LLMs and Generative AI

Short-Term Research Projects and Proof of Concepts

Common Applications

AI Model Training
(Robotics, Image Generation)

High-performance GPUs facilitate rapid development of models requiring substantial data and computation.

LLM Foundation
Model Development

Supports the extensive computational resources needed for training large language models.

Short-Term Research Projects
in Academia and Industry

Offers cost-effective and flexible GPU resources for time-limited projects.

Dedicated
Generative AI Infrastructure

Provides secure, dedicated GPU environments for in-house AI development.

PLAN

AI SPACON Cloud Pricing Plans*

Specification	Single Node	Cluster Configuration
Specs	HGX H200 × 1 Node	HGX H200 × Multiple Nodes
Monthly Fee	USD17,692.31	Contact Us
GPU / Node	NVIDIA H200 (SXM) × 8	NVIDIA H200 (SXM) × 8
GPU Memory / Node	Total 1128GB	Total 1128GB
GPU Memory Bandwidth	4.8TB/s	4.8TB/s
GPU Interconnect	NVLink・NVSwitch	NVLink・NVSwitch
vCPU / Node	112 cores	112 cores
System Memory / Node	2TB	2TB
Storage / Node	30TB	30TB
Performance / Node (FP16 Tensor Core)	15.8PFLOPS	15.8PFLOPS
Number of Nodes	1 Node	Multiple Nodes (customizable)
Inter-Node Bandwidth	-	1600Gbps
OS	Ubuntu OS	Ubuntu OS
Region	Japan	Japan

PLAN

Discount plans for NVIDIA HGX H200*

Years	Monthly Fee
1-Year Contract	USD 15,923.08 (10% off)
2-Year Contract	USD 15,038.46 (15% off)
3-Year Contract	USD 11,323.08 (36% off)
5-Year Contract	USD 10,792.31 (about 39% off)

* Based on an exchange rate of USD 1 = JPY 143.
* Actual rates vary depending on the current exchange rate at the time of signing.

Security and Support

About GPUSOROBAN

GPUSOROBAN is an AI development-focused GPU cloud service provided by Highreso Co., Ltd. It offers supercomputer-level computational performance in a convenient cloud environment, enabling rapid training and execution of complex AI models.

Real-World Use CasesN

GPUSOROBAN：AI SPACON Cloud