What Singapore’s New AI Data-Centre Capacity Means for Facility Managers

Practical lessons on cooling, power quality, sensor coverage and maintenance readiness for Singapore facilities

Singapore’s growing AI infrastructure is putting renewed attention on the engineering systems that keep critical facilities operating. On 21 August 2026, EDB and IMDA announced the provisional allocation of 200 MW of new data-centre capacity to four operators under Singapore’s second data-centre call for applications.

For data-centre operators, this expansion brings intense focus to cooling, electrical resilience, energy monitoring and digital operations. For facility managers outside the data-centre sector, it offers a useful case study: reliability is not created by one advanced technology. It comes from understanding critical loads, monitoring the right conditions, planning shutdowns and turning information into timely action.

Warehouses, industrial buildings, offices and SMEs do not need data-centre-scale infrastructure to apply these lessons. They can begin with a clearer view of risk and a targeted digital maintenance plan.

Why AI-intensive facilities raise the engineering challenge

AI workloads can place demanding and variable loads on information-technology equipment. That increases the importance of thermal management and stable electrical supply. Cooling systems, pumps, fans, power-distribution equipment, backup systems and controls must work together rather than operate as isolated assets.

The lesson for other facilities is straightforward: identify which equipment is operationally critical, understand what conditions can cause failure and make those conditions visible before they become an incident.

This does not mean every warehouse needs complex data-centre cooling. It does mean that a temperature-sensitive storage area, automated material-handling system, server room, production line or building-management system should have an appropriate monitoring and response strategy.

1. Start with a critical-load and heat map

Facility teams should map where electrical demand, heat and business impact are concentrated. The map should include more than the main incoming supply. Consider distribution boards, transformers, UPS systems where present, refrigeration or ACMV equipment, server and communications rooms, automation systems, dock equipment and fire or life-safety-related services.

For each area, record:

• What the equipment supports and how critical it is.
• What happens if it fails for 15 minutes, one hour or longer.
• Which environmental conditions can affect performance.
• Whether there is standby capacity, manual fallback or an approved recovery procedure.
• Who receives the alert and who has authority to respond.

This exercise helps separate genuinely critical assets from equipment that can be repaired during normal working hours. It also supports better maintenance budgets and more realistic continuity planning.

2. Treat cooling as a reliability system, not just a comfort system

In many facilities, cooling is viewed mainly through occupant comfort or energy consumption. In AI-intensive and equipment-heavy environments, it is also a reliability control. A failed fan, blocked filter, faulty sensor, leaking pipe, overloaded circuit or poorly maintained pump can create a rapid operational problem.

Facility managers should review whether temperature, humidity and equipment status are monitored in the locations that matter most. Monitoring may include supply and return temperatures, room conditions, motor or pump status, differential pressure, water leakage and abnormal runtime. The exact sensor set should be based on the facility’s equipment and risk profile.

Preventive maintenance should also verify that sensors are positioned correctly and that alarms are meaningful. An alarm that is too sensitive creates fatigue; an alarm that is too late provides little protection.

3. Look beyond power availability to power quality

A facility may have sufficient electrical capacity and still experience problems caused by poor power quality or weak changeover planning. Voltage disturbances, harmonics, unbalanced loads, loose connections, overheating components and poorly coordinated protection can affect sensitive equipment and automation systems.

A practical review can include:

• Thermal inspection of distribution equipment where appropriate.
• Load and phase-balance checks.
• Review of recurring trips, nuisance alarms and unexplained equipment resets.
• Condition checks on batteries, generators, switchboards and critical controls where installed.
• Verification of automatic and manual changeover procedures.
• Clear records of which loads may be shed during an incident.

Any electrical testing or work should be planned and carried out by suitably qualified personnel. The management priority is to ensure that findings are documented, assigned and closed rather than left as informal observations.

4. Build useful sensor coverage before adding advanced AI

BCA’s Smart Facilities Management and Artificial Intelligence for the Built Environment resources point towards data-driven operations, workflow automation and AI-supported diagnostics. These capabilities depend on usable operational data.

Before investing in advanced analytics, check whether basic information is available, consistent and connected to the people responsible for action. Useful starting points may include temperature, humidity, energy consumption, vibration, equipment runtime, differential pressure, water leakage, access events and maintenance status.

Singapore’s edge-AI research activity also highlights the potential for more intelligence to be processed closer to equipment and sensors. In practical terms, local or near-device analytics may support faster alerts where connectivity, response time or data volume is a concern. However, the technology should match the operational need. A simple threshold alert that reaches the right person can be more valuable than a complex model that nobody trusts or acts on.

5. Turn monitoring into a maintenance workflow

Condition monitoring is only useful when it leads to a controlled response. Every important alert should have an owner, a severity level, an expected response time and a documented action.

A basic workflow might be:

1. Detect an abnormal reading or equipment state.
2. Validate the alert against operating conditions and recent maintenance.
3. Assign the issue to the responsible technician or service provider.
4. Record the inspection, finding, action and parts used.
5. Escalate if the issue affects a critical load or repeats.
6. Review recurring events for a permanent corrective action.

This is where AI automation and digital services can provide practical value. Automated notifications, work-order creation, escalation rules, digital checklists and maintenance dashboards can reduce delays and improve traceability without requiring a fully autonomous facility.

6. Strengthen shutdown and permit-to-work planning

Maintenance readiness includes knowing how work will be carried out safely. Before planned shutdowns, teams should confirm affected systems, isolation points, backup arrangements, restart sequences, communication channels and acceptance checks.

MOM identifies electronic permit-to-work systems, IoT environmental sensors, digital alerts, video analytics and other technologies as relevant to facilities management, logistics and industrial workplaces. For facility operators, a digital permit-to-work process can help make approvals, isolations, risk controls and close-out records easier to review.

Technology does not replace competent supervision or safe work procedures. It helps make the process more visible and consistent, particularly when multiple contractors or operating teams are involved.

A practical starting plan for Singapore facilities

Facility managers can begin with a 30-day readiness review:

• List the top ten assets whose failure would disrupt operations.
• Identify current cooling, electrical and environmental readings.
• Check whether alerts reach a named person outside office hours.
• Review the last three unplanned failures or recurring alarms.
• Document one controlled shutdown and restart procedure.
• Choose one maintenance workflow to digitise and measure.

The goal is not to copy a data-centre design. It is to adopt the discipline behind high-reliability operations: critical-load visibility, dependable cooling, power-quality awareness, actionable sensors and maintenance processes that support fast, safe decisions.

How ISS can help

ISS supports businesses exploring engineering, facility-management and AI automation requirements. Depending on the operating environment, this may include reviewing monitoring needs, structuring digital maintenance workflows, improving alert and escalation logic, and identifying practical opportunities for automation.

Contact ISS to discuss how your facility can improve cooling oversight, power resilience, maintenance readiness and operational visibility without overcomplicating the solution.

Sources: EDB and IMDA data-centre allocation announcement; Singapore edge-AI research activity; BCA Artificial Intelligence for the Built Environment; BCA Smart Facilities Management; MOM Workplace Safety and Health Technology.