OPERATIONS
Keep it running, day and night.
We help you run AI campuses as critical infrastructure: availability, efficiency and security, measured continuously.
SERVICES
What we take on
Operation and maintenance
Preventive and corrective maintenance, spare parts, subcontractors and procedures: every system stays within its operating envelope.
Colocation and compute capacity
Space, power and cooling ready for GPU clusters, or compute capacity made available.
24/7 monitoring
Power, cooling and security followed around the clock, with escalation and incident management procedures.
Energy, carbon and water
Measure to reduce: energy efficiency and carbon intensity, with regular reporting, and proof, month after month, that no water is consumed.
Physical security and compliance
Access control, video surveillance, procedures and audits, to the requirements of clients and authorities.
OPERATIONS
A loop that never stops
Monitor
Sensors, alarms and rounds: see it before it breaks.
Maintain
Planned work, spares at hand, proven procedures.
Optimise
Match cooling and power to the actual load.
Report
Indicators shared with the client, with no black box.
INDICATORS
What we measure
- Availability
- The share of time the promised capacity is actually delivered.
- PUE
- The campus’s total energy divided by the energy of its IT equipment.
- WUE
- The water consumed for each unit of IT energy: zero on our campuses, cooled in a closed loop.
- Carbon
- The emissions tied to the energy consumed, tracked then reduced.
- Recovery
- The time it takes to return to normal after an incident.
IN PLAIN TERMS
Frequently asked questions
What does operating an AI data centre cover?
Preventive and corrective maintenance, monitoring of power, cooling and security day and night, incident management and reporting to the client. The aim: that every piece of equipment stays within its operating range.
What is PUE, and why track it?
PUE compares the total energy of the campus with that of its IT equipment. The lower it is, the less energy the campus spends on anything other than compute. Tracking it continuously makes it possible to adjust cooling to the actual load.
How do you prove that a campus consumes no water?
By measuring. WUE compares the water consumed with the IT energy: on a campus cooled in a closed loop, it is zero. We record it and report it month after month, along with energy efficiency and carbon.
What happens when an incident occurs?
Alarms trigger tested escalation procedures: teams step in with parts at hand and written instructions. The time back to normal is measured, and every incident feeds preventive maintenance.