40HC Modular AI
Data Center
Deploy high-density AI compute in a factory-integrated modular AI data center with cooling, power, fire protection, monitoring and service access in one transportable 40HC module.
40HC
Factory-built module
Liquid or Air
Cooling architectures
Up to 300 KW
Reference liquid input
-25°C to 45°C
Reference environment
Turn electrical power into
AI compute, then move the heat outside
Turn electrical power into
AI compute, then move the heat outside
system. GPU heat is captured at the server, managed through the CDU, rejected outdoors and returned as a stable cooling loop.
Power reaches the GPU racks
Project power enters the main distribution system and is delivered to GPU servers, cooling equipment and auxiliary controls.
Server heat enters the cooling loop
Cold plates and server manifolds capture high-density GPU heat and carry it into the CDU-controlled secondary loop.
Heat leaves the module
The CDU maintains flow, pressure and temperature while the dry cooler or chiller rejects heat outdoors. DCIM or BMS monitors the complete operating state.
1、Power & Compute
Power Distribution & GPU Racks
Electrical power is distributed to the GPU servers and converted into AI compute and heat.
2、Heat Capture
Cold plates / Server manifolds
Cold plates cool GPUs while manifolds circulate coolant across racks.
3、Loop Control
CDU
The CDU isolates server and facility loops while managing heat exchange and fluid parameters.
4、Heat Rejection
Dry cooler / Chiller
The outdoor system transfers captured server heat to ambient air at the project operating point.
5、Monitoring & Return
DCIM / BMS / Cooled supply
Coolant returns to GPUs while monitoring tracks system metrics, leaks, and alarms.
One factory-built platform
Seven coordinated infrastructure systems
A reliable AI deployment starts by matching the server load to the enclosure, cooling loop, power architecture and safety systems. ACT integrates them before shipment so the site team receives one coordinated module.
01 - COMPUTE RACKS
Customized server spatial layout, engineered weight distribution for floor-load safety, and dedicated service clearance for unhindered maintenance.
02 - COOLING - CDU
Please ensure the proper CDU, heat rejection system, or precision air cooling equipment is selected around the specific thermal load.
03 - POWER DISTRIBUTION
Please confirm the main electrical input feed, circuit protection, power metering, PDU system, and grounding configuration.
04 - FIRE PROTECTION
Please integrate custom project-engineered fire detection with automated clean-agent fire suppression protection for the module.
05 - MONITORING
06 - SERVICE ACCESS
Please design convenient front, rear, and side access clearance for seamless hardware maintenance and overall equipment replacement.
07 - OUTDOOR ENCLOSURE
Please review and configure weatherproofing, storm drainage, freeze protection, and anti-corrosion treatment options according to specific site requirements.
Select the cooling architecture
around the AI load
Select the cooling architecture
around the AI load
Cooling selection should follow the server heat split, required inlet conditions, site climate and availability target.
HIGH-DENSITY REFERENCE ARCHITECTURE
Transfer server heat through a controlled CDU-based secondary loop
A distributed manifold system supports controlled flow, front-and-rear service access and project-defined pump and cooling-equipment redundancy.
Server capacity
35 servers
System flow
31.5 m³/h
Per-server flow
15 LPM
Supply range
15-40°C
Operating pressure
1-4 bar
Rack groups
5 groups
A distributed manifold system supports controlled flow, front-and-rear service access and project-defined pump and cooling-equipment redundancy.
CONTAINED-AISLE REFERENCE ARCHITECTURE
Coordinate rack density with precision cooling and outdoor condensers
A contained hot-and-cold-aisle layout controls server inlet conditions, while optional cooling-pad pre-cooling can be evaluated for suitable climates and water conditions.
Rack quantity
5 racks
Rack capacity
46U each
Total rack space
230U
Indoor units
4 × CR066
Outdoor units
2 × LVC170
Nominal reference
264 KW
Reference configuration. Usable rack space and IT load depend on server dimensions, power, airflow, cable routing and maintenance clearance.
Build the approval pathway
into the module design
Build the approval pathway
into the module design
The final electrical, fire-protection and inspection pathway depends on the site, utility, equipment list and local Authority Having Jurisdiction.
Power Architecture
Voltage, protection, PDU, metering and grounding selected around the final server load.
Power Architecture
Listed or recognized components can be selected according to the approved project bill of materials.
Fire Protection
NFPA-referenced, project-engineered clean-agent protection based on the protected volume.
Detection and Release
Smoke and heat detection with alarm stages, release delay, manual release and emergency stop.
Monitoring Integration
Project-specific data integration interfaces tailored for DCIM, BMS or the customer’s existing central monitoring platform.
Field Coordination
Ensure comprehensive planning around the local AHJ, utility company requirements, and agreed field-inspection scope.
Designed around the equipment
technicians must actually maintain
Rack quantity is only one capacity limit. The layout must also protect cable routes, fluid connections, emergency access and component replacement paths.
Service access
Sliding and side maintenance doors support replacement and inspection tasks.
Fluid interfaces
Pipe connections, cooling valves, fill and drain points remain serviceable.
Electrical separation
Strong-current and low-voltage routes are separated for maintenance clarity.
Outdoor interfaces
Rain protection is applied around intake, exhaust, power and sensor locations.
Emergency access
Emergency exits and escape hardware are included in the layout.
Grounding
Two distributed grounding points support the final site grounding design.
From server list
to commissioned module
From server list
to commissioned module
A defined engineering sequence protects capacity, cooling and compliance decisions before fabrication starts.
Workload Review
Please review and confirm the exact server model, total unit quantity, physical dimensions, power consumption, and thermal load split.
Site Review
Please kindly confirm the site climate conditions, local altitude, water source, operational voltage, concrete foundation, and delivery logistics.
System Engineering
Please accurately size the equipment racks, cooling capacity, power requirements, fire suppression systems, controls, and maintenance clearances.
Design Approval
Please carefully review and validate the complete design drawings, module interfaces, itemized bill of materials, and master project schedule.
Factory Integration
Please carefully assemble the integrated module and conduct thorough mechanical verification, complete electrical testing, and control system checks.
Delivery Support
Please efficiently coordinate overall site transport, modular placement, electrical connections, initial system startup, and final client handover.
40HC modular AI data center
reference configurations
Final parameters are confirmed in the project technical agreement. Values below remain tied to their specific reference architecture.
| Parameter | Liquid-cooled reference | Air-cooled reference | Engineering note |
| Container format | 40HC standard marine container | 12,192 × 2,438 × 2,896 mm | Structural details follow the approved project arrangement. |
| Server / rack reference | 35 high-density AI servers | Five 46U racks, 230U total | Capacity must be checked against dimensions, weight, power and service clearance. |
| Server power versions | 6.1 kW per server / approximately 7 kW per server | Server-specific | The two liquid values represent separate source versions and are not combined. |
| IT load versions | 213.5 kW / approximately 225 kW | 238 kW modeled IT load | 264 kW is the air system’s nominal cooling reference, not the modeled IT load. |
| Container input | Approximately 300 kW reference input | Project-specific | The liquid reference includes approximately 75 kW of auxiliary load. |
| Cooling equipment | CDU plus project heat-rejection system | 4 × Vertiv CR066 and 2 × LVC170 | Equipment selection is confirmed against the operating point. |
| System flow | 31.5 m³/h total; 15 LPM per server | Not applicable | Final flow follows server pressure-drop and cold-plate requirements. |
| Supply temperature | 15-40°C reference range | Server inlet temperature controlled by air system | Actual setpoint follows server and site requirements. |
| Pressure | 1-4 bar server operating range; 4 bar system maximum | Not applicable | Maximum design pressure is not the normal operating pressure. |
| Coolant | Water and ethylene-glycol mixture | Refrigerant and air system by equipment selection | Glycol concentration follows freeze protection and compatibility requirements. |
| Electrical references | AC 240V/60Hz system input; AC 220V, 50/60Hz server-side source reference | 415V/480V, three-phase, 50/60Hz | Final input and load-side architecture must be confirmed per project. |
| PDU interfaces | Server-specific | 56 × C19; 36 × C13; 14 × three-phase aviation connectors | Interface quantity and standard can be adjusted to the selected server. |
| Environment | -25°C to 45°C; 8%-90% RH reference | -25°C to 45°C reference | Site extremes, altitude, dust, salt fog and water conditions require review. |
| Redundancy | Dual-main-pump concept; 3+1 or N+1 cooling options | Project-specific | Final redundancy follows the required availability target. |
| Fire protection | Project-engineered clean-agent system with smoke and heat detection | North American reference design uses an NFPA-referenced approach. | |
| Monitoring | Temperature, leak, smoke and electrical monitoring with project interfaces | DCIM or BMS integration is defined in the project scope. | |
FAQ
Procurement questions to resolve before the quotation.
Capacity depends on server dimensions, rack units, weight, power, cooling method, cable routing and service clearance. The liquid reference uses 35 high-density AI servers, while the air reference provides five 46U racks.
Liquid cooling is generally preferred when server heat density or liquid-to-air heat split exceeds the practical capacity of the air system. Selection should follow a heat-load and operating-point review.
ACT needs the server model and quantity, rated and peak power, dimensions, cooling interface, site temperature and altitude, available voltage, redundancy target, location and schedule.
North American voltage, protection, PDU and component options can be engineered around the project load and approved bill of materials. The final pathway depends on the site and local AHJ.
No. Component listings or recognitions do not automatically certify the complete container. The certification and inspection scope must be confirmed for the project.
PUE depends on the IT load, ambient temperature, cooling architecture, supply temperature, pump and fan operation, heat-rejection equipment and redundancy mode.
Yes. Multi-module projects can be planned around shared heat-rejection, electrical and monitoring infrastructure. Interfaces and redundancy should be defined during initial site design.
Commissioning support can be defined according to the delivery scope, project location, contractor responsibilities and startup requirements.
Selected Modular Infrastructure Projects
Crypto LLC Phase I Tennessee, USA
North American compliant 40ft high-cube modular mining containers with custom PDU and integrated safety systems.
QRB Labs 80MW Project Ethiopia
High-density air-cooled mining container cluster with advanced dust and sand protection.
200MW Hyperscale Hydro-Cooling | Norway
Ultra-large closed-loop water-cooled facility optimized for maximum efficiency and low PUE.
Multiple Global Deployments
Over 500MW of mining containers successfully delivered across three continents.
Request a Modular AI Data Center Configuration Review
Server Model、IT Load、Rack Quantity、Rack Density、Project Location、Available Voltage、Cooling Preference、Redundancy Target、Deployment Schedule