We offer more than just a GPU cloud service. As a "growth engine" that paves the way for the future, we provide a next-generation IT infrastructure that supports the very foundations of industry. Highreso is committed to exponentially increasing the world's overall computing power, providing steadfast support to all researchers and companies dedicated to pushing the boundaries of innovation.
High-performance GPU cloud services designed to accelerate intensive computing tasks, including Generative AI and Large Language Models (LLMs). By optimizing data center construction and operational efficiency, we deliver premium GPU server power at a competitive price point.
NVIDIA B200-powered clusters, billed per GPU, per minute. No cluster engineering required.
HGX H200 instances with unified GPU memory — tuned for LLM and multimodal workloads.
Production-grade NVIDIA GPU instances for inference, generative AI and ML pipelines.
Designed from silicon to data center to keep cost, latency and sovereignty on your side.
Vertically integrated data centers and per-minute billing eliminate the hyperscaler margin stack — without trading away performance.
Four domestic data centers in Ishikawa, Takamatsu and Genkai, certified to ISO 27001 / 27017 and Privacy Mark standards.
Every account gets direct access to GPU specialists — whether you are training a 70B model or onboarding your first team.
No procurement, no quotas, no capacity requests. Select a GPU, launch, and pay only for the minutes you use.
We own and operate purpose-built, GPU-first data centers across Japan — engineered for power density, thermal efficiency and regional resilience.
We partner with local governments, universities and industry leaders to open GPU data centers that revitalize regional economies — turning former school buildings, renewable power and domestic engineering into world-class AI infrastructure.
Every HIGHRESO data center is rooted in its community. We create new jobs, repurpose closed school buildings and run digital-literacy programs — from smartphone classes to AI learning sessions — in partnership with local governments in Ishikawa, Kagawa and Saga.
Our Shika (Ishikawa) facility uses a hybrid cooling design that combines air conditioning with 100% outside-air cooling, and our sites integrate in-house renewable energy installations to keep operations energy-efficient and low-carbon.
Service, servers and data centers are all designed in-house. That vertical integration is how we deliver over 50% lower pricing than AWS or GCP — with no data-egress fees and flat monthly plans available for predictable budgeting.
Japanese-language engineering support, domestic contracting, and yen-denominated pricing — shielding your GPU budget from FX swings and delivering the kind of technical partnership global hyperscalers can't match inside Japan.
State-of-the-art GPU data center built inside a previously underutilized research facility, deployed under METI's "Cloud Program" supply-assurance plan with ¥7.7 billion of public support. GPUSOROBAN service has been running here since December 2024.
| Facility | Repurposed research building |
|---|---|
| Services | AI Supercomputer Cloud · Compute Cluster · Private Cloud |
| GPU | NVIDIA H200 (ultra-high-end) |
| Online | December 2024 |
One of Japan's largest dedicated GPU data centers, in continuous operation since 2019. A hybrid cooling design combines air-conditioning rooms with 100% outside-air cooling rooms for high efficiency. Local hiring helps address depopulation, ageing and labor shortages while creating new industries and jobs in the region.
| Facility | Hybrid AC + 100% outside-air cooling |
|---|---|
| Services | High-Performance Computing · Remote Workstation · Image-generation SaaS |
| GPU | NVIDIA A100 · A4000 |
| Online | 2019 (DC 1) · 2022 (DC 2) |
Closed rural school building reborn as a GPU compute hub. The site distributes GPU workload from our Shika facility, complementing core capacity while contributing to local revitalization in Genkai-cho.
| Facility | Repurposed school · Shika DC compute distribution hub |
|---|---|
| Services | Image-generation SaaS |
| GPU | NVIDIA A4000 |
| Online | August 2025 |
Closed school building converted into an ultra-high-end GPU environment, designed to serve the most demanding AI training and inference workloads on next-generation NVIDIA hardware.
| Facility | Repurposed school |
|---|---|
| Services | AI Supercomputer Cloud · Compute Cluster · Private Cloud |
| GPU | NVIDIA H200 (ultra-high-end) |
| Online | 2026 |
Founded in 2007, HIGHRESO Co., Ltd. is a Japanese technology company on a mission to accelerate global innovation through computing power. Our 120+ team designs, builds and operates one of the country's largest GPU cloud infrastructures.
3-24-1 Ichigaya-Tamachi, Shinjuku-ku, Tokyo, Japan
106-1 Moroura, Genkai-cho, Higashimatsuura-gun, Saga, Japan
NVIDIA · Hokuriku Electric Power · Toyota Tsusho · NTT PC Communications · Tech Wind · Cresco · Applied · SeedWave · NICT (National Institute of Information and Communications Technology) · JAIST (Japan Advanced Institute of Science and Technology) · and many more.
Tell us about your workload — pilot access, enterprise contracts or custom capacity.