Modular data center solutions

We design & manufacture mobile HPC infrastructure.

Deploy high-density compute capacity in 12 to 16 weeks instead of years. From HPC containers to transformers, BESS, generators, cooling and GPU clusters: a complete modular data center, ready to operate anywhere in the world.

Talk to an engineer

12–16

weeks, order to operation

1.5 MW

per modular ISO container

Full stack: compute, power, cooling & monitoring
Optional BESS and diesel generator integration
Turn-key: from MV connection to GPU-ready racks
Ideal for AI clusters, HPC, rendering & simulation
Designed in alignment with EN 50600
Suited for NVIDIA & AMD

A fully modular data center, delivered turn-key

Every part built off-site, pre-tested, plug-and-play.

This drastically reduces engineering time, civil works and deployment risk.

HPC containers

Rear Door Heat Exchanger (RDHx)
Direct Liquid Cooled (DLC)
Immersion cooled
Suited for NVIDIA & AMD

Power modules

LV switchgear
PDU's
Busbars

Cooling modules

Closed-loop dry coolers
Pump modules
Heat-reuse hot-water loop

Energy modules

Battery Energy Storage Systems (BESS)
Diesel generator back-up

MV infrastructure

MV switchgear
Transformers

Compute

GPU clusters (NVIDIA & AMD)
Pre-configured racks

Why modular?

Faster, flexible and lower cost.

Deploy in weeks

Traditional data centers take a minimum of 36 months to be operational. Our modular solution deploys in 12 to 16 weeks, including power and cooling infrastructure.

Scale as you grow

Add modular blocks of power or compute as the need increases. No wasted capex or overbuilding. Every component can be relocated.

Lower risk

All modules are pre-engineered and validated at our factory. On-site work is minimised, which shortens the timeline and improves reliability.

One HPC container. Three cooling technologies.

Full HPC solution →

Immersion cooling. Servers submerged in dielectric fluid. Ultra-high density per tank, lowest PUE.

Direct liquid cooling. Coolant to the cold plates of GPU and CPU. Outlet heat suited for a hot-water loop.

Rear door heat exchanger. 55–200 kW per rack for conventional air-cooled servers, no chillers.

Power and cooling infrastructure

The modules around the compute container: heat rejection, medium-voltage power and on-site generation, all containerised and pre-tested.

Infrastructure →

External or roof-mounted closed-loop dry coolers; hot-water loop for heat reuse

MV switchgear & transformers, containerised

Gas-to-power generation for off-grid sites

Other specifications

A+B power feed
Integrated LV panel
Integrated PLC
Monitored cooling loop
High efficiency dry cooler
Optional hot-water loop reuse
Customer customisations by request

Monitor and control

Beckhoff PLC
Modbus, SNMP & REST API
Power metering on main busbar, rack & outlet
Cooling loop telemetry (flow, pressure & temperature)
Generator / BESS integration
Cloud dashboard for remote operations: our CubeApp (CubeBridge on the edge, CubeTwin in the cloud)

CubeApp · included with every Datacube

Every container has a brain on the edge and a twin in the cloud.

The CubeBridge, an industrial edge server inside each container, controls it autonomously and keeps working during internet loss. It synchronises with CubeTwin, the cloud portal where operators and clients monitor and control the fleet.

Live sensors: climate, airflow, electrical, rack temperatures, motor drives
Rack PDUs, network, cameras and alarms with AI incident summaries
No inbound ports: outbound HTTPS only, mandatory 2FA, every action logged
AI advises, AI never controls: the control path is deterministic, on the edge

Container overview

A container in one view, with live camera frames.

Status, power and rack temperatures next to the latest frame from every camera on site: hall, yard and gate. Every module the container carries shows as a chip.

Sensors

Climate, airflow, safety, electrical.

Cold and warm aisle, intake and exhaust airflow, door and smoke, voltage, current and power factor. Air-conditioning, rack probes and motor drives read out over Modbus.

Rack PDUs

Power down to the outlet.

Per-PDU and per-outlet metering over SNMPv3, derived outlet watts, outlet history with two-tier retention and switching with a 15-second PSU rest time.

Network

The site network as telemetry.

Switches, access points, tunnels and clients from the site controller; devices follow their MAC, so a replaced unit needs no reconfiguration.

Alarms

Thresholds are configuration, not code.

Severity, hysteresis and repeat interval per alarm and per container, pushed to the edge. Dashboard, e-mail and Telegram as channels. Critical alarms pinned on top.

AI incident summary

Analyse an alarm on request.

What happened, when and the likely correlation, with recommended first steps and the sources used. Written only when a person asks; it never enters a notification.

History

Minute data to yearly trends.

Power, temperature, humidity and uptime over 24 hours to 12 months, resolution chosen automatically, events as clickable markers, two periods as an overlay.

Copilot

Ask the fleet a question.

Fleet-wide answers grounded in read-only queries of live telemetry and alarm history, sources listed under every answer. It cannot start or stop anything.

Reports & exports

CSV and Excel, as a background job.

Choose containers, period and data types; the Export Center shows progress and keeps files for 30 days. Every export is logged.

Activity

Every action, traceable.

Commands, grid signals, configuration changes, exports and AI calls in one feed: who, when, why, previous and new state.

Tower

The control room wall.

A dark wall display for the whole fleet with a live event feed, built for a screen that is on all day.

Cameras

Eyes on hall, yard and gate.

Latest frame per camera through the site tunnel, with motion and people events in the alarm stream. Recorded footage stays in the camera system.

Mobile

The same platform, on site.

A PWA down to 360 px with the fleet, live container, alarms and copilot, so operators on site work from the same truth.

Why institutions choose Datacubes

Built to specification. Delivered to standard.

Designed in alignment with EN 50600, factory pre-tested before shipping
Schneider & Siemens components, Beckhoff PLC, Modbus / SNMP / REST API
Power metering on busbar, rack and outlet; cooling loop telemetry
A+B power feed, optional BESS and generator integration
Installed and commissioned on site by our own engineers

Standards

EN 50600

CSC / UL

The Datacubes ecosystem

From energy source to useful heat.

Power in, compute out, heat reused. Each step is a containerised module we design and manufacture ourselves.

Energy sourceGrid, renewables, BESS, generators or Datacube Gas

Power conversionMV switchgear, transformers, LV panels and busbars

ComputeHPC containers with RDHx, DLC or immersion cooling

Heat reuseHot-water loop for district heating, greenhouses, industry

For energy suppliers

Put unused power capacity to work.

HPC workloads offer a stable, flexible demand for electricity. Place a modular data center at the source and turn surplus or stranded energy into revenue, without investing in data center infrastructure.

Energy partnerships →

Grid stabilisation

Absorb overproduction from renewable sources.

Energy parks

Convert energy at wind or solar farms that would otherwise be lost.

Remote locations

Immediate returns from energy at sites far from demand.

Heat reuse

Waste heat for district heating and industrial processes.