ENTERPRISE AI
Infrastructure for your own AI workloads
Coordinate the facility with the systems your team will operate. Define the rack configuration, network and storage, workload requirements and operating plan before freezing the design.
MDC3000 · AMD HELIOS INFRASTRUCTURE
A modular facility for eight AMD Helios racks.
MDC3000 brings 2.75 MW of target IT capacity into a prefabricated modular building. Power, liquid cooling and service space are coordinated for eight AMD Helios compute racks and 16 racks for networking, CPU and storage.
- 2.75 MW
- Target IT load
- 3.2 MW
- Total incoming power, current design
- 8
- AMD Helios compute racks
- 16
- Network, CPU and storage racks
Current design figures. Final power, cooling and rack schedules are coordinated with the selected OEM equipment and site conditions. IT load and total incoming power are separate design quantities.
Power and cooling sized for the same cluster.
Rack-scale AI needs more than space for compute racks. Power distribution, liquid connections, network and storage equipment, residual air cooling and maintenance access have to work together.
MDC3000 coordinates these systems around AMD Helios in one prefabricated modular facility, for enterprise AI, inference services and training workloads.
Four systems designed together.
- 01
Power subassembly
Incoming power, UPS, batteries and rack distribution coordinated for the Helios equipment schedule. The current design calls for 3.2 MW of total incoming power; upstream equipment and supply boundaries are agreed in the project scope.
- 02
IT environment
Eight wide-format AMD Helios compute racks and 16 network, CPU and storage racks. The IT-hall access openings are adapted for the wider compute equipment and its installation and service requirements.
- 03
Thermal & facility subassembly
Liquid-cooling distribution, pumping, controls, leak monitoring and supporting air cooling, with fire-protection and monitoring interfaces coordinated for the project.
- 04
Heat rejection
Outdoor heat rejection selected for the climate and availability requirements. The design starts with dry heat rejection and provides for adiabatic assistance where conditions require it.
The building retains the external dimensions and appearance of the existing modular platform. The Helios rack layout, access openings, transport splits and service clearances are detailed for the selected equipment.
8 compute racks. 16 for networking, CPU and storage.
The Helios configuration separates wide compute racks from the supporting equipment. All 24 rack positions are included in the coordinated facility layout.
| Rack allocation | Quantity | Design role |
|---|---|---|
| AMD Helios compute racks | 8 | Wide-format rack-scale compute equipment |
| Network, CPU and storage racks | 16 | Cluster networking, CPU services and storage |
| Total racks | 24 | Compute and supporting IT equipment |
The 2.75 MW IT-load target covers the complete IT configuration. It is not a per-rack rating: the final rack schedule allocates power between compute and supporting equipment.
Plan the facility around your Helios deployment.
AI SERVICE OPERATORS
Capacity for inference and training services
Plan compute, supporting IT, cooling and available site power together. Align equipment procurement, facility delivery and commissioning with the capacity you intend to bring into service.
The selected OEM system sets the rack dimensions, weight, power connections, cooling interfaces and service requirements. Equipment availability and the delivery programme are agreed for the project.
Direct liquid cooling for the main load. Air for the rest.
The current design direction is approximately 85% direct liquid cooling and 15% air cooling. Liquid cooling serves the main compute load; air cooling removes residual heat and heat from supporting equipment.
85%Direct liquid cooling in the current design direction
15%Air cooling for residual and supporting loads
- Compute cold plates
- OEM cooling interfaces
- Facility cooling loop
- Site-selected heat rejection
- Residual rack + network / CPU / storage heat
- Air-side cooling
- Site heat rejection
Coolant specification, CDU arrangement, flow rates and operating temperatures are coordinated with the selected Helios OEM system.
The 85% / 15% split is an approximate design allocation. The project thermal schedule establishes the actual liquid and air duties.
Match heat rejection to the climate.
The facility cooling loop recirculates coolant. Dry heat rejection is the starting strategy, with adiabatic assistance where the climate requires it.
Annual energy and water use depend on the location, IT-load profile, equipment and cooling controls. The project model establishes the expected PUE, water demand and any storage or refill requirements.
2.75 MW for IT. 3.2 MW total incoming power.
The current MDC3000 design separates power for compute and supporting IT from the total incoming supply, which also serves cooling, electrical losses and facility systems.
The final electrical study confirms the connection, UPS arrangement, rack feeds, maintenance modes and redundancy for the selected hardware and site.
Annual energy performance is modelled for the site climate, operating load and cooling configuration.
Design the site around the cluster.
- 01
Confirm the equipment and location.
Define the Helios OEM configuration, network, storage, IT load, site climate and deployment plan.
- 02
Approve the basis of design.
Freeze electrical, thermal, civil, controls, fire and OEM boundaries.
- 03
Complete detailed engineering.
Review wide-rack access openings, installation routes, fibre routes, rack power, liquid interfaces, transport and outdoor heat rejection.
- 04
Factory-integrate the AI Block.
Assemble power, thermal, controls and monitoring systems in a controlled environment.
- 05
Complete factory acceptance testing (FAT).
Test the agreed electrical, thermal, control and alarm functions before shipment.
- 06
Install and connect.
Complete site works in parallel, then position and integrate the system.
- 07
Complete SAT / IST and commissioning.
Verify the transported and site-connected facility before operational handover.
The delivery programme separates factory work, transport, site installation and commissioning, with the dependencies for each stage agreed in the proposal.
Current design figures and project scope
- Target IT load
- 2.75 MW
- Total incoming power, current design
- 3.2 MW
- Compute racks
- 8 AMD Helios racks
- Supporting IT racks
- 16 for networking, CPU and storage
- Total rack positions
- 24
- Purpose
- Enterprise AI, inference and training
- Cooling direction
- Approximately 85% direct liquid / 15% air
- External form
- Same external dimensions and appearance as the existing modular platform
- IT-hall access
- Openings adapted for wide-format Helios equipment
- OEM interfaces
- Rack feeds, coolant, CDU arrangement and network interfaces coordinated with the selected OEM
- Heat rejection
- Dry-first strategy; adiabatic assistance where required by climate
- Project scope
- Infrastructure, computing equipment and site works defined separately
- Acceptance
- Agreed FAT, site acceptance, integrated testing and commissioning
These figures describe the current MDC3000 design. The project basis of design confirms the equipment, operating conditions, power and thermal schedules, redundancy and scope of supply.
Agree the infrastructure and equipment scope.
INSIDE THE COORDINATED MDC3000 DESIGN
- IT environment, wide-rack access and service clearances
- LV receiving interface, UPS, batteries and rack distribution
- Facility cooling-loop distribution and air-side support cooling
- Controls, monitoring and leak detection
- Coordinated fire-protection, security and BMS/DCIM interfaces
- Factory integration and the agreed FAT scope
CONFIRMED BY PROJECT
- Compute, network and storage hardware
- OEM rack supply and CDU / cooling-loop implementation
- Upstream MV switchgear and transformers
- Standby generation
- External heat-rejection equipment and redundancy
- Site civil works, foundations, utilities and access
- Fire detection/suppression equipment and local approvals
- SAT, integrated testing, certification and operational support
FOR SITE OWNERS AND ENERGY COMPANIES
Build capacity for a tenant’s computing equipment.
MDC3000 can provide the facility for a tenant operating AMD Helios equipment. Coordinate the power, cooling and service space with the tenant’s cluster requirements, and discuss potential tenant introductions where relevant.
Tenant introductions and commercial terms are project-specific; occupancy and investment returns are not guaranteed.
Frequently asked questions
What is MDC3000 designed for?
MDC3000 is a prefabricated modular data-center design for eight AMD Helios compute racks and 16 racks for networking, CPU and storage. Its current target is 2.75 MW of IT load, for enterprise AI, inference and training deployments.
Does 2.75 MW include the network, CPU and storage equipment?
The 2.75 MW target is the overall IT load for the selected configuration, including compute and supporting IT. The final equipment schedule assigns the load to individual racks. It is separate from the current 3.2 MW total incoming-power design figure.
How is the equipment cooled?
The current design direction is approximately 85% direct liquid cooling and 15% air cooling. The selected OEM equipment and final thermal schedule establish the liquid and air duties, coolant requirements, CDU arrangement and operating temperatures.
What building and access requirements should I plan for?
The external dimensions and appearance follow the existing modular platform. IT-hall access openings are adapted for the wider Helios racks. We issue the configuration-specific layout, foundation loads, service clearances, delivery route and lifting requirements during project engineering.
What PUE and water use should I expect?
Energy and water performance are modelled for the site climate, IT-load profile and cooling configuration. Dry heat rejection is the starting approach, with adiabatic assistance where required. A closed facility loop does not by itself establish zero water consumption.
Are the computing systems included?
The core offer is the data-center infrastructure. The proposal states who supplies the Helios compute racks, networking, CPU and storage equipment, and where the OEM cooling interfaces sit. Those systems are not automatically included in the infrastructure package.
What availability and certification requirements apply?
The project basis of design defines redundancy, maintenance modes, local requirements and acceptance tests. Any formal certification or OEM qualification must identify its assessed configuration and scope.
How much does MDC3000 cost and how long does delivery take?
The quotation follows the equipment configuration, site, cooling and resilience requirements and supply boundaries. The programme separates engineering, procurement, factory assembly and testing, transport, installation and commissioning.
Can MDC3000 support a colocation project?
MDC3000 can provide infrastructure for a tenant operating Helios equipment. The facility design follows the tenant’s hardware and operating requirements. For suitable projects, ModulEdge can discuss potential tenant introductions; occupancy and investment returns are not guaranteed.
Plan the facility for your Helios cluster.
Share your hardware plan, site location and available power. We will discuss the rack layout, infrastructure scope and delivery programme.