Designed for
AI density.
NNDC I / NNDC II · Data-centre facilities
From utility power to usable compute.
An AI facility must solve power, heat, networking and operations together. NNDC takes a liquid-cooling-first engineering approach to dedicated racks, cages, halls and offtaker-led deployments.
| Facility | Development position | Discussion focus |
|---|---|---|
| NNDC I | 20 MW development programme with an offtaker secured. | Phasing, workload and delivery scope under definitive agreements. |
| NNDC II | 20 MW utility capacity; TNB power letter obtained; target completion late 2027 to mid-2028. | Dedicated AI facilities, powered-shell development, long-term offtake and colocation partnerships. |
Project status is supplied by NNDC and subject to project data-room evidence, definitive agreements and delivery conditions.
Cooling matched to the workload.
| Approach | Suitable workloads | Design considerations |
|---|---|---|
| Direct-to-chip liquid cooling | High-density GPU training and inference | CDUs, secondary loops, water chemistry, leak detection and rack compatibility. |
| Rear-door heat exchangers | Mixed-density halls and transition deployments | Residual air heat, plumbing and service access. |
| Chilled-water / air cooling | General-purpose servers, networking and storage | Containment, airflow and design ambient conditions. |
| Immersion cooling | Hardware-qualified specialist deployments | Material compatibility, warranty, maintenance and fire review. |
A clear capacity basis.
IT capacity = utility power ÷ PUE. For example, 20 MW ÷ 1.25 = 16 MW of IT load. This is a design illustration, not a certified PUE or saleable-capacity commitment. Server counts also depend on rack density, equipment draw, networking, storage and operating headroom.
Resilience
Tier III design basis targeting concurrent maintainability. Enhanced redundancy can be evaluated; Uptime Institute has no formal “Tier 3.5” classification.
Explore ↗Energy
Evaluate onsite and contracted renewables, storage and backup generation. Annual renewable share and backup autonomy are assessed separately.
Explore ↗Readiness
Confirm grid, water, fibre routes, land and approval conditions before committing the delivery programme.
Explore ↗Data-centre operations for AI.
Deployment readiness, liquid-cooling management, infrastructure monitoring, maintenance and incident response—with defined responsibilities and customer reporting.
Explore DC Operations ↗