AI Data Centers
Crypto Infrastructure
AI Data Centers
Crypto Infrastructure
CoolBase-1080A/W is a pre-engineered, outdoor modular AI data center in a standard 40HQ form factor. Air-cooled (A) or hydro-cooled (W) on the same platform — deploy with either mode, migrate between them without re-architecting. 1 MW of IT load per block.
CoolBase-1080A/W consolidates power, cooling, IT cabinet, fire protection, and DCIM operations into a pre-engineered 40HQ cluster. It could be installed on any flat outdoor site, connected to medium-voltage grids, and brought online in weeks— rather than the 18-24 months required for conventional construction. We support customization for air cooling or hydro cooling to match your project requirements, and all settings can be confirmed before production starts. CoolBase-1080A/W is built on a unified hardware platform with air and hydro cooling options, compatible with most server models. We will define the suitable cooling configuration for you based on your compute density roadmap.The hydro-cooled variant uses CDU and rack manifolds to deliver coolant to cold plates; the air-cooled variant is equipped with integrated CRAC precision air conditioning units for heat dissipation. Built-in CRAC units maintain constant ambient inside the pod. Air cooling is ideal for moderate-density GPU deployments, retrofit projects, and sites where liquid infrastructure is not yet ready.
Compatible servers:
Air-cooled configurations — supports B200 / B300 series and other air-cooled server platforms.
Integrated CDU with primary/secondary dual-loop architecture. Direct-to-chip cold plates deliver high-capacity thermal management for current and next-generation accelerator platforms.
Compatible servers:
Hydro-cooled configurations — supports GB200 / GB300 and other liquid cold-plate server platforms. Note: Cold-plate liquid cooling only. Immersion cooling is not within this product's scope.
Key infrastructure advantages. Each key subsystem — CDU, power distribution, UPS, generator, and cooling system — adopts 2N redundancy. Select modules include N+1 internal redundancy, delivering system availability above 99.99%. Supports medium-voltage high-voltage input for direct grid connection, simplifying the power supply chain for high-density computing loads. Site prep and module manufacturing run in parallel, not in sequence. Modules arrive pre-tested and ready to connect. Compress your time-to-power by half. 40HQ standard container, weather-sealed, dust-resistant, and ambient-rated for hot climates and remote locations. No traditional data center building required. IT pod features flexible thermal management compatibility, supporting air-cooled, hydro-cooled, and hybrid configurations. One pane of glass for power, thermal, security, and SLA. Intelligent alerting, automated reports, and remote O&M reduce on-site staffing by an order of magnitude. Every CoolBase-1080A/W ships as a complete kit. The IT pod, cooling plant, power train, and operations cabin are pre-built, pre-tested, and pre-commissioned at our facility before they ever touch your site. 40-foot high-cube ISO container housing 8 high-density server cabinets, integrated CDU, and built-in CRAC for dual-mode cooling. Two 1080 kW cooling plants running in 2N configuration. Each station includes dry coolers, pumps, and a primary-side heat-rejection loop. Solid-state-transformer or medium-voltage module accepts high-voltage grid input directly, eliminating the need for a separate substation build. Battery-backed UPS modules ride through the gap between utility interruption and generator response, keeping the IT load on clean power. Two diesel generator modules with automatic transfer switch, sized for 24-hour full-load operation including belly tank and acoustic enclosure. A dedicated standalone test bay for periodic on-load generator testing, enabling performance validation without disrupting IT operations. Built-in circuit protection receives upstream power to supply IT pods, CDUs, cooling stations and auxiliary equipment for continuous and stable computing operation. On-site operations center running CoolSpace's DCIM platform. Real-time visibility into power, thermal, security, and SLA across the entire cluster. Site work and module manufacturing run in parallel. You don't wait for one to finish before the other starts. The result: a complete, tested AIDC infrastructure block online in 9 months, vs. 18+ for a traditional build. Our engineering team plans your site around your compute, cooling, and density profile. Soil, grid, fiber, and access are all validated up front. Utility interconnection, ground prep, and access run in parallel with module manufacturing and factory acceptance testing at our facility. Pre-tested modules are delivered, set in position, and connected. The IT pod, cooling plant, power train, and O&M cabin are commissioned together. Power flows, cooling connects, and the site is rack-ready. From grid connection to first AI workload in days, not quarters. CoolBase-1080A/W is the building block. Each unit is one megawatt of IT load, fully self-contained, and designed to cluster. Place three blocks for 3 MW. Six for 6 MW. Scale linearly to the size of your AI factory — without redesigning anything. The 40HQ pod fits eight 52U server cabinets, each rated for 135 kW of IT load. That's enough headroom for NVIDIA GB300 NVL72-class hardware, with N+1 redundancy on the cooling distribution unit overhead. Each cabinet feeds into a shared CDU with 2N internal pumps, so a single pump or manifold failure never reaches the rack. The result is a pod that runs at full load, 24/7, in continuous AI training duty cycles — and keeps running through fault events. The hydro-cooled configuration uses a primary-secondary dual-loop architecture. The primary loop carries heat to the outdoor cooling station, where dry coolers and pumps reject it to ambient. The secondary loop stays inside the IT pod, isolated from the outside environment, with the CDU acting as the heat exchange hub between the two. Blue cold supply enters the rack manifolds at controlled temperature and flow. The liquid absorbs chip heat and returns warm to the CDU, where it transfers that energy to the primary loop and is sent back cold. The system is leak-monitored, flow-balanced, and temperature-stable even under full load transients. Outdoor cooling station · heat rejection to ambient · wide temperature range. Inside the IT pod · CDU-mediated · closed and monitored · feeds rack cold plates. Pre-engineered, pre-tested, and pre-documented. Detailed mechanical, electrical, and thermal drawings available under NDA. Whether you're planning a new AI factory, upgrading an existing facility, or scaling a multi-megawatt deployment — share your compute, density, and grid profile. Our engineers will return a CoolBase-1080A/W configuration, footprint, and timeline in days.
From bare site to AI workload.
9 months.A complete AI data center. One 40HQ block.
Dual Thermal Architecture — Air and Hydro on One Platform
Air-Cooled Configuration
Hydro-Cooled Configuration
Engineered for Continuous Operation
Redundancy Architecture
Medium-Voltage Direct Input
9-month deployment
Outdoor, harsh-environment rated
Dual Cooling Architecture
DCIM centralized operations
Integrated Modules. One Coordinated System.
IT Pod 40HQ
Cooling Station Outdoor
SST / MV Power Module 10 kV
UPS Power Module N+1
Diesel Generator ATS
Generator Testing Pod
Smart Power Distribution Cabin
O&M Control Pod DCIM
Parallel-Path Deployment — 9 Months to Energization
Site Design
Site Prep & Manufacturing
Modules Placed
Energization
Modular Scalability — 1 MW to Gigawatt-Class AI Factories
Inside the IT Pod — 8 × 135 kW Cabinets, 1 MW Total IT Load
Primary-Secondary Dual-Loop Cooling Architecture
Primary Loop
Secondary Loop
CoolBase-1080A/W at a glance.
Share your compute requirements. Our engineers will return a CoolBase-1080A/W configuration within days.