AI in Power: How Machine Learning Is Driving Smarter, Cleaner, and More Reliable Data Centres in Singapore

AI Powering the Future of Singapore’s Data Centres

Amid the rapid digital acceleration, Singapore stands as a leading data centre hub in Asia, supporting everything from cloud computing and AI services to the region’s digital transformation. However, this position comes with pressing energy demands. In 2023, global data centre energy consumption soared to approximately 240–340TWh—about 1–1.3% of worldwide electricity use (Mytton, 2025), with expectations to double between 2022 and 2026 as AI workloads surge. Singapore’s own Green Data Centre Roadmap sets out to expand capacity by an ambitious 300MW, with a clear directive: every new megawatt must align with advanced energy efficiency and sustainability goals (IMDA, 2024; Rajah & Tann Asia , 2024).

As AI systems become exponentially more complex, the power required to train and deploy models such as GPT-4 or advanced generative AI grows larger and more resource-hungry. For example, the US saw data centre power use triple over the past decade, peaking at 176TWh in 2023—4.4% of its total electricity use (Shehabi, et al., 2024), with GPU-powered AI servers jumping from under 2TWh in 2017 to over 40TWh by 2023 (Shehabi, et al., 2024). This escalating trend is mirrored in Asia, where Singapore’s humid climate makes cooling even more challenging and energy-intensive than in temperate regions.

Yet, AI is not simply raising these challenges—it is also unlocking solutions.

When deployed smartly, machine learning and predictive analytics can optimise cooling cycles, forecast energy demand, and automate capacity planning to avoid overprovisioning and energy wastage (Bernard, et al., 2025).

Through Singapore’s strategic initiatives—focusing on green energy, advanced cooling innovations, and intelligent IT management—the nation aims to turn its data centres into exemplars of sustainable infrastructure. As we dive deeper into the intersection of AI, energy management and sustainability, we find the potential to transform data centres from consumption giants into smart, efficient, and environmentally aligned engines of growth.

Contents

• How AI Optimises Energy Use in Singapore’s Data Centres
• The Sustainability Impact
• The Future of AI-Enabled Data Centres
• A Smarter Path to AI-Ready Data Centres

How AI Optimises Energy Use in Singapore’s Data Centres

Artificial intelligence and machine learning are spearheading a seismic shift in how energy is consumed and optimised within modern data centres, particularly in high-growth digital economies like Singapore. As data centre operations become ever more complex—driven by mounting computing loads from AI, cloud, and edge workloads—the imperative for smarter, data-driven energy management is unprecedented.

Intelligent Cooling: Biggest Source of Energy Savings

Cooling systems account for up to 40% of a data centre’s total electricity usage, making them a prime target for optimisation. Singapore’s humid climate amplifies this challenge, but AI solutions are rising to meet it:

• AI-powered cooling technologies (such as those pioneered by KoolLogix in partnership with A*STAR) have reduced cooling energy needs by as much as 50% in regional installations (A*STAR, n.d.). These systems use real-time sensor data and predictive algorithms to dynamically adjust cooling setpoints, saving electricity, reducing carbon emissions, and protecting equipment from overheating—all without manual intervention.
• Leading global operators—including Google and Huawei—have demonstrated that machine learning can deliver sustained energy reductions. Google’s DeepMind-driven platform slashed data centre cooling energy by 40% (Evans & Gao, 2016; Safrina & Kusno, 2024), while Huawei’s iCooling@AI saw up to a PUE improvement of 8-15% in field tests across large data centres that commercially deployed iCooling@AI (Fei & Song, 2020).

Singapore’s Green Data Centre Roadmap sets world-class benchmarks, requiring new builds to achieve a Power Usage Effectiveness (PUE) rating of 1.3 or lower—an efficiency standard significantly tighter than regional averages (typically 1.55–1.6) (Safrina & Kusno, 2024). AI makes hitting these targets possible by automating granular controls and forecasting conditions in real time.

Predictive Maintenance and Capacity Planning

• Machine learning reduces downtime and waste by constantly monitoring for anomalies in energy draw, equipment temperatures, and operational performance. Automated alerts and maintenance scheduling prevent costly failures and unnecessary usage spikes, supporting 24/7 uptime and regulatory compliance (North, 2025).
• Predictive analytics support smarter capacity planning, reducing overprovisioning—historically a major source of wasted power—by matching resource availability more closely to actual demand.

Demand Forecasting and Grid Integration

• AI solutions forecast variable workload peaks and troughs, adjusting energy use and cooling to match predicted needs with maximum precision. These systems help align local and regional grid requirements, particularly important as Singapore integrates more renewable energy into its infrastructure (Mordor Intelligence, 2025).
• As generative AI fuels unprecedented growth, global data centre electricity consumption is forecast to double from 536TWh in 2025 to over 1,065TWh by 2030. Advanced AI management mitigates these increases, keeping business energy costs lower and minimising environmental impact (Ramachandran, et al., 2024).

The Sustainability Impact: AI’s Role in Reducing Carbon Footprint and Enabling Greener Data Centres

AI-enabled energy optimisation is becoming central to Singapore’s efforts to enhance the sustainability of its burgeoning data centre sector. As electricity demand rises—driven by digitalisation and AI proliferation—so too does the need for measurable reductions in energy consumption and greenhouse gas emissions.

Singapore’s Green Standards and Targets

Singapore’s Green Data Centre Roadmap sets clear requirements for the industry: 300MW of incremental data centre capacity must integrate advanced energy-saving measures, and future expansion hinges on the deployment of green energy. This is complemented by the refreshed Green Mark for Data Centres 2024 (GMDC:2024), a world-first sustainability standard, which benchmarks facilities on operational efficiency, renewable energy adoption, and carbon footprint reduction.

To achieve and verify these gains, data centre operators are leveraging digital twins and AI-driven simulations to generate carbon footprint reports, supporting Green Mark certification even before facilities are constructed. These digital tools allow predictive modelling of energy flows and sustainability outcomes—a process increasingly recognised as essential for greenfield and brownfield projects alike (IMDA, 2025).

AI’s Measurable Influence on Energy and Emissions

• Dynamic optimisation by AI enables real-time adjustments to energy-intensive operations, from cooling to workload distribution. Singapore’s solutions have demonstrated up to 50% lower cooling-related energy usage, reducing sector carbon emissions significantly (IMDA, 2025).
• Globally, operators such as Meta have cut operational emissions by up to 94% through full adoption of green energy for data centres, aided by AI-powered resource management platforms (Pieyre, 2025).
• Advanced machine learning reduces overprovisioning and avoids waste, with predictive maintenance algorithms supporting reductions in unnecessary energy draw and hardware replacements.

Strategic Advantages and Environmental Alignment

Singapore’s drive for green, AI-managed data centres is not just about sustainability; it confers competitive and reputational benefits. Organisations adopting “Green AI” experience not only cost savings, but also enhanced operational efficiency and brand perception as eco-leaders in the region (IMDA, 2025).

Strikingly, the Green Data Centre Standard (SS 564) enables companies to continually assess and improve their energy profile, embedding sustainability into daily operations and technology procurement decisions.

The Future of AI-Enabled Data Centres: Towards Autonomy and Resilience

Singapore’s next-generation data centres are rapidly evolving from static infrastructure into intelligent ecosystems, underpinned by artificial intelligence, automation, and sustainability as core design tenets. With the digital economy booming and demand for data processing surging, tomorrow’s data centre is expected to be both smarter and greener—delivering digital growth while minimising environmental impact (Sharon, 2025).

Autonomous Operations Through AI

AI-driven optimisation is now foundational in state-of-the-art data centres. Advanced facilities such as Nxera’s upcoming Tuas development in Singapore are already integrating intelligent cooling, automated resource management, and real-time energy optimisation, using machine learning and predictive analytics to adjust energy consumption dynamically. This allows for proactive management—dynamically balancing loads, circumventing inefficiencies, and maintaining uptime in the face of fluctuating demand and environmental conditions (Sharon, 2025).

Looking ahead, the future is autonomous: AI’s deep learning models, fed by complex sensor data fusion, will enable self-adjusting systems that improve themselves continuously, detect and resolve anomalies in real time, and dynamically scale resources without human intervention. Digital twins will play a key role, allowing operators to simulate various physical and operational scenarios virtually for fast, low-risk decision-making (Eco, 2025).

Enabling Scalability and Resilience

Singapore’s AI-optimised data centres are incorporating high-performance hardware (like NVIDIA DGX-Ready certified facilities) and solid-state power electronics—such as those developed through ABB and DG Matrix’s collaboration—which achieve energy efficiencies as high as 98%. These innovations ensure that expanding workloads, including rapid AI adoption and cloud migration, can be handled seamlessly, with minimal environmental footprint and operational risk (Mordor Intelligence, 2025).

This trajectory is reinforced by national strategies including the Green Data Centre Roadmap and Green Mark for Data Centres 2024, which push operators to adopt smart energy systems, real-time monitoring, and low-impact operations, setting global benchmarks for scale and sustainability (Sharon, 2025).

A Smarter Path to AI-Ready Data Centres

As Singapore solidifies its position as a top-tier data centre hub, the pressure to maximize uptime and efficiency is immense. Many operators believe the answer lies in complex, expensive AI platforms, but this often skips the most critical step. The biggest gains in reliability and cost savings come from getting the fundamentals right first.

The good news is that you can achieve significant performance wins by leveraging the data and controls you already own. By focusing on clean data signals, calibrated controls, and smarter alarm management, you build a solid "AI-ready" foundation that delivers immediate value and prepares you for any future analytics, on your own terms.

To provide a clear path forward, we have created our new guidebook:

“AI-READY UPTIME: PREDICT • OPTIMIZE • PERFORM.” This is not a theoretical paper; it’s a practical toolkit containing:

• The AI Readiness Scorecard: A 15-minute self-check to score your facility's data, infrastructure, and culture.
• The ROI Blueprint: A framework to build a solid business case, showing how getting the basics right saves money now.
• The 3-Step Playbook: A hands-on action plan to Predict, Optimize, and Perform using your existing systems—no new software required.

Download the guide today to take the first practical step toward reducing risk and waste. Move forward with evidence, not hype.

DOWNLOAD

References

1. A*STAR. (n.d.). Transforming Data Centre Efficiency: KoolLogix’s Sustainable Cooling Solutions with A*STAR. Retrieved from A*STAR: https://www.a-star.edu.sg/enterprise/our-stories/digital-services-solutions/transforming-data-centre-efficiency-koolLogixs-sustainable-cooling-solutions-with-astar
2. Bernard, R. N., de Lange, R., Lakshmanan, S., & Likens, S. (2025). Could net-zero AI become a reality? Retrieved from PwC: https://www.pwc.com/gx/en/issues/value-in-motion/ai-energy-consumption-net-zero.html
3. Eco. (2025). Artificial Intelligence Meets Data Centre – How AI is Changing the Industry. Retrieved from Eco: https://international.eco.de/news/artificial-intelligence-meets-data-centre-how-ai-is-changing-the-industry/
4. Evans, R., & Gao, J. (2016). DeepMind AI reduces energy used for cooling Google data centers by 40%. Retrieved from Google: https://blog.google/outreach-initiatives/environment/deepmind-ai-reduces-energy-used-for/
5. Fei, Z., & Song, X. (2020). iCooling@AI: Smart cooling for data centers. Retrieved from Huawei: https://www.huawei.com/en/huaweitech/publication/90/smart-cooling-data-centers
6. IMDA. (2024). Singapore Green Data Centre Roadmap. Retrieved from IMDA: https://www.imda.gov.sg/-/media/imda/files/how-we-can-help/green-dc-roadmap/green-dc-roadmap.pdf
7. IMDA. (2025). The role of Green AI in Singapore’s digital sustainability journey. Retrieved from IMDA: https://www.imda.gov.sg/resources/blog/blog-articles/2024/12/role-green-ai-in-sg-digital-sustainability-journey
8. IMDA. (2025). Turning the red dot, green: Helping data centres get better at staying cool. Retrieved from IMDA: https://www.imda.gov.sg/resources/blog/blog-articles/2025/02/red-dot-analytics-help-data-centres-be-cool
9. Mordor Intelligence. (2025). Singapore Artificial Intelligence (AI) Optimised Data Center Market Size & Share Analysis - Growth Trends & Forecasts (2025 - 2030). Retrieved from Mordor Intelligence: https://www.mordorintelligence.com/industry-reports/singapore-artificial-intelligence-ai-data-center-market
10. Mytton, D. (2025). Data center energy and AI in 2025. Retrieved from /dev/sustainability: https://www.devsustainability.com/p/data-center-energy-and-ai-in-2025
11. North, M. (2025). Here's how data centre heat can warm your home. Retrieved from World Economic Forum: https://www.weforum.org/stories/2025/06/sustainable-data-centre-heating/
12. Pieyre, L. (2025). AI, Data Centres, and Their Expanding Impact on the Environment. Retrieved from Solar Impulse Foundation: https://solarimpulse.com/news/ai-data-centres-and-their-expanding-impact-on-the-environment#
13. Rajah & Tann Asia . (2024). Singapore’s Green Data Centre Roadmap – Carving a Route for a Sustainable Digital Future. Retrieved from Rajah & Tann Asia : https://www.rajahtannasia.com/viewpoints/singapores-green-data-centre-roadmap-carving-a-route-for-a-sustainable-digital-future/
14. Ramachandran, K., Stewart, D., Hardin, K., Crossan, G., & Bucaille, A. (2024). As generative AI asks for more power, data centers seek more reliable, cleaner energy solutions. Retrieved from Deloitte: https://www.deloitte.com/us/en/insights/industry/technology/technology-media-and-telecom-predictions/2025/genai-power-consumption-creates-need-for-more-sustainable-data-centers.html
15. Safrina, R., & Kusno, M. B. (2024). Harnessing Artificial Intelligence for a Greener Data Centre. Retrieved from ASEAN Centre for Energy: https://aseanenergy.org/post/harnessing-artificial-intelligence-for-a-greener-data-center/
16. Sharon, A. (2025). Singapore: Smarter, Greener Data Centres for a Sustainable Future. Retrieved from OpenGov: https://opengovasia.com/singapore-smarter-greener-data-centres-for-a-sustainable-future/?c=sg
17. Shehabi, A., Newkirk, A., Smith, S. J., Hubbard, A., Lei, N., Siddik, M., . . . Sartor, D. (2024). 2024 United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory. doi:doi.org/10.71468/P1WC7Q
18. Singapore Standards EShop. (n.d.). SS 564-1:2020. Retrieved from Singapore Standards EShop: https://www.singaporestandardseshop.sg/Product/SSPdtDetail/ac609aae-e97c-456f-a7a1-5258a2816b45

• power systems
• Green data centre
• AI
• machine learning
• cooling optimisation
• singapore
• Energy Efficiency
• sustainability
• data centres
• Predictive maintenance