August DataCentre

Data centres are the backbone of today's digital economy, powering everything from cloud computing to generative AI. As demand for digital services continues to rise, global data centre capacity, and its electricity consumption, is also growing rapidly. According to International Energy Agency (IEA) reports, global data centre electricity demand recently surged by 17% and is projected to double by 20301. However, this increase in computing power generates significant heat, making reliable cooling essential to protect equipment, prevent downtime and maintain operations.

With the rise of generative AI and high-performance computing (HPC), cooling has evolved from an auxiliary utility into a business-critical function in data centre design for financial performance and environmental compliance. Modern heating, ventilation and air conditioning (HVAC) systems now play a vital role in ensuring operational reliability and optimising energy efficiency.

The evolution of HVAC deployment
High-Density Thermal Challenges
Physics Limit
Air cooling becomes ineffective beyond ~30 kW per rack due to airflow limitations causing heat density.
The Threat
If an operator cannot cool a 100-kW rack, they cannot run AI hardware, making the facility obsolete with power strain and reliability risks due to overheating.
Safeguarding Uptime and Operational Reliability
Data centres demand up to 99.999% uptime, with less than six minutes of downtime per year. Effective thermal management is critical to preventing unplanned outages3. Strategic HVAC moves operators from reactive fixing to continuous uptime.

Past: Traditional Air Cooling

Standard racks used 5kW to 10kW, easily cooled by traditional air systems (raised floors, CRACs, and aisle containment).

Current: Demand and Limitation

Modern AI and GPU clusters have significantly increased power demands, pushing rack densities from 50 kW to over 100 kW. Traditional air cooling reaches its physical limit at approximately 30 kW per rack, as the volume of air required to dissipate such high heat loads becomes unsustainable.

Energy Efficiency & Financial Performance

Data centres consume an estimated 1–2% of total global electricity4, with HVAC infrastructure using 30–40% of a facility's total power. As the largest non-computing energy consumer, HVAC majorly impacts operating costs.

Cooling efficiency is measured by Power Usage Effectiveness (PUE). Legacy data centres typically operate at a PUE of 1.5–2.0, where nearly every watt powering IT equipment requires another watt for cooling and power distribution5.

By upgrading to advanced HVAC systems with advanced controls and variable-speed drives (VSD), large-scale facilities consuming 100 megawatts (MW) of power can achieve PUEs below 1.26, significantly reducing energy consumption and yielding millions of dollars in annual savings.
Transitioning to Liquid Cooling
Modern HVAC architecture is shifting away from traditional air-cooling toward highly efficient liquid cooling solutions. With 25 times higher thermal conductivity than air, liquid is an exceptionally efficient cooling medium for high-density environments7.
Key cooling approaches:
Direct-to-Chip Cooling: Uses closed loops to pipe a dielectric fluid or treated water to a metal cold plate on the CPU or GPU. The liquid absorbs heat and pumps it to an external heat exchanger, cooling racks up to 100 kW.
Hybrid Layouts: Combines air cooling for standard compute rows with liquid-loop infrastructure, supporting localised AI deployments.
Stay Cool. We’ve got the data covered.

The Reality: Growing data demands and denser hardware are pushing traditional cooling to its limits, making smarter HVAC essential.

The Competitive Edge: HVAC now defines a data centre's performance, uptime, and financial returns. Advanced cooling technologies, such as liquid cooling, AI, automation and sustainable heat rejection, secures a clear competitive edge.

Discover how advanced cooling integrates into modern data centres and LAPP solutions built to maximise uptime in high-density environments.