Elite-class Enterprise CDUs — 1.38 MW to 2.6 MW with full redundancy. The workhorse of every DCX-certified AI training cluster.
Capacity: 1.38 MW or 2.6 MW. Every unit is DCX-engineered in Europe, certified for 45°C warm-water operation, and delivered with Middle East-grade corrosion protection. Backed by our Dubai and Doha-based engineering teams for design, installation and 24/7 service.
| Parameter | Value |
|---|---|
| Cooling capacity | 1380 / 2600 kW |
| Form factor | Rack-based, 42U / 47U variants |
| Primary water inlet | Up to 45°C |
| Redundancy | N+1 or 2N |
| Service access | Front + rear, hot-swap modular |
| Pumps | EC variable-speed, hot-swappable |
| Controls | 10" HMI + remote web UI |
| Integration | SNMP, BACnet, Modbus, REST API |
| Power | 400 V / 3-phase / 50 Hz dual feed |
| Lead time to GCC | 12–16 weeks |
NVIDIA HGX / DGX rows running 24/7
National research labs, weather centres, genomics
UAE and KSA sovereign compute programs
Premium colocation offering liquid-cooled suites
Reservoir modelling, seismic analysis
HFT clusters and risk modelling
We can configure N+1, 2N or concurrent-maintainable depending on your uptime SLA. Most hyperscale AI customers pick 2N.
Yes — we plumb a single secondary loop that feeds both cold-plate and immersion racks. Temperature balance is maintained by zone controllers.
Yes. We ship with Modbus, BACnet and SNMP native. We also build custom integrations for proprietary DCIMs.
The 2600 runs at ~72 dB at 1m at full load — quieter than most 400 kW air CRACs.
Yes. Our local team sees every telemetry point 24/7 via DCX's cloud telemetry platform. Predictive-failure alerts trigger automatically.
