Singapore’s Proposed Digital Infrastructure Bill: A Practical Preparation Guide for Data-Centre Facility Teams

Prepare your power, cooling, energy and operational data for stronger resilience and future compliance expectations.

Singapore’s proposed Digital Infrastructure Bill is expected to bring greater attention to how major data centres manage resilience, operational incidents and energy efficiency. For facility managers, engineering contractors and data-centre owners, the practical question is not only what the Bill may require, but whether existing systems can produce reliable operational evidence when needed.

The Ministry of Digital Development and Information introduced the Bill on 8 September 2026. Based on the current announcement, the proposed framework would cover major data centres and address resilience risks involving power, cooling, fire and other physical or operational incidents. Data centres with at least 3MW of critical IT load would also be expected to meet facility-level energy-efficiency requirements, including power usage effectiveness, or PUE, requirements.

The framework is proposed legislation, not an enacted requirement. Detailed implementation arrangements, technical requirements and transition provisions remain subject to further engagement. The checklist below therefore separates what has been publicly indicated from ISS recommendations for preparing an AI-ready facility operation.

What the proposed framework means for facility teams

The direction is clear: large digital infrastructure sites will need stronger control over the physical systems that support computing capacity. This includes electrical distribution, standby power, cooling, fire protection, monitoring, maintenance and business continuity.

For operators, resilience should not be treated as a document prepared only for an audit or incident review. It should be visible in daily operating data. Teams should be able to identify abnormal conditions, understand the affected assets, escalate incidents quickly and show what action was taken.

Current public direction to note

• The proposed Bill would establish licensing regimes for major data centres and address risks to digital infrastructure resilience.
• Relevant risks may include power, cooling, fire and other physical or operational incidents.
• Incident-reporting obligations are part of the proposed framework.
• Data centres with at least 3MW of critical IT load would be subject to facility-level energy-efficiency requirements, including PUE-related requirements.
• Detailed implementation and transition arrangements are not yet finalised based on the current announcement.

These points should not be read as a complete technical compliance checklist. Operators should monitor official updates and obtain professional advice for site-specific obligations.

Practical preparation checklist

1. Establish a reliable asset and systems register

Start with a current register covering electrical, mechanical, fire and building systems. Include transformers, switchboards, UPS systems, generators, chillers, cooling towers, pumps, CRAH or CRAC units, sensors, meters, fire-protection equipment and critical control panels.

The register should identify asset location, function, capacity, redundancy arrangement, maintenance responsibility, last service date and operational criticality. A simple spreadsheet may be a starting point, but larger facilities will benefit from a structured asset database that can connect to building-management, electrical-monitoring and maintenance systems.

This information supports incident reporting, planned maintenance, root-cause analysis and continuity planning. It also gives AI or analytics tools a more dependable foundation.

2. Improve power and cooling visibility

Facility teams should review whether they can see the operating condition of critical power and cooling assets at a useful level of detail. Depending on the site, this may include electrical current, voltage, load, power factor, breaker status, UPS condition, generator readiness, temperature, humidity, chilled-water conditions, airflow and equipment run status.

Monitoring should be designed around operational decisions, not data collection alone. A temperature reading is more useful when it is linked to the affected room, rack zone, cooling unit, alarm limit and responsible response team. Similarly, an electrical alarm should identify the relevant distribution path and the possible impact on critical loads.

Wireless or wired sensing may be appropriate depending on the asset, environment, installation constraints and cybersecurity requirements. The key objective is dependable, time-stamped data that can be reviewed during normal operation and after an event.

3. Build a PUE data workflow

PUE is calculated by comparing total facility energy with the energy used by IT equipment. The calculation is only as useful as the boundaries, meters and data quality behind it.

Operators should document which meters support the calculation, how often readings are captured, how missing data is handled and who reviews unusual changes. Where possible, energy data should be separated by meaningful loads such as IT equipment, cooling, pumps, lighting and other facility services.

Do not wait until a reporting deadline to discover that meters are poorly located, inconsistent or not synchronised. A baseline period can help teams understand normal operating patterns and investigate changes caused by weather, utilisation, cooling settings, maintenance or equipment faults.

4. Strengthen incident reporting and escalation

A practical incident workflow should cover detection, verification, classification, escalation, containment, recovery and post-incident review. It should define who is contacted for power, cooling, fire, water ingress, control-system failure and other operational events.

Alarm floods can delay response, so teams should review alarm priorities and remove duplicate or non-actionable notifications where appropriate. Each high-priority alarm should have a clear response instruction, owner and escalation time. Incident records should capture the event time, affected asset, symptoms, action taken, service impact and follow-up requirements.

This creates a more reliable operational record and can support future reporting obligations, subject to the final requirements of the Bill.

5. Test business continuity, not only emergency documents

Business-continuity plans should be tested through realistic scenarios. Examples include loss of utility power, generator failure, cooling degradation, chilled-water interruption, fire-system activation, networked monitoring failure and restricted access to a plant room.

Exercises should include facility operators, engineering contractors, security, IT operations, building management and relevant business stakeholders. The goal is to identify practical gaps: unclear authority, unavailable spare parts, incomplete contact lists, inaccessible drawings, delayed vendor response or insufficient manual operating procedures.

After each exercise, record corrective actions and assign owners. A continuity plan that is not linked to maintenance, training and operational data will quickly become outdated.

6. Move from calendar maintenance towards condition-based decisions

Preventive maintenance remains important, but condition data can help teams prioritise work between scheduled service intervals. Trends such as rising temperature, increased vibration, abnormal current, repeated trips, declining battery performance or changes in chilled-water behaviour may indicate developing problems.

Analytics can help identify patterns and prioritise inspection. However, AI should support engineering judgement rather than replace it. Before using predictive models, confirm that sensors are calibrated, data is consistent, failure events are recorded and maintenance teams understand how recommendations will be used.

A sensible starting point is a focused use case, such as monitoring critical cooling or electrical assets, validating alert quality and measuring whether the workflow reduces nuisance alarms or improves response time.

How to make a facility AI-ready

AI readiness is primarily an information and workflow challenge. A facility may have many sensors but still lack usable data if asset names differ between systems, timestamps are unreliable, alarms are not prioritised or maintenance records are incomplete.

ISS recommends reviewing five foundations:

1. Data quality: consistent asset IDs, timestamps, units and measurement locations.
2. System integration: practical links between BMS, electrical monitoring, energy meters, CMMS and incident records.
3. Operational ownership: clear responsibility for reviewing alerts and approving action.
4. Cybersecurity and access control: controlled access, appropriate segmentation and documented data flows.
5. Human validation: engineering review before automated recommendations become operating decisions.

These are ISS recommendations for implementation, not stated requirements that AI systems must be deployed under the proposed Bill.

A sensible next step for Singapore operators

Facility teams can begin with a readiness review covering critical assets, sensor coverage, meter boundaries, PUE data, alarm workflows, maintenance history and continuity exercises. The review should identify immediate gaps, low-disruption improvements and longer-term integration opportunities.

For data-centre owners, FM providers, warehouse operators and building owners, the same principles apply: reliable operational data, clear response workflows and measurable energy performance support both resilience and better day-to-day decisions.

Singapore’s proposed Digital Infrastructure Bill may evolve as further details are developed. Preparing the underlying engineering data and operating discipline now can help organisations respond more confidently to future requirements while improving visibility, maintenance planning and energy management today.

Contact ISS to discuss engineering, facility management or AI automation requirements for your site. Visit intelligencesolutionservice.com.