_Behind the Cloud: The Hidden Economics of Data Centre FM Efficiency

India's data-centre story is usually told in very large numbers. Capacity. Megawatts. Capital expenditure. Artificial intelligence. Cloud migration. Hyperscale campuses. There is good reason for the excitement.

India’s operational data-centre capacity reached approximately 1.6 GW (1,600 MW) by the end of 2025, according to Knight Frank Research, representing a more than fivefold increase from 296 MW in 2016. Knight Frank estimates that the total upcoming data-centre development pipeline across India’s major data-centre markets stands at 8.33 GW (8,326.6 MW).

Mumbai ranks 15th globally in operational data-centre capacity, with 766 MW of live IT capacity, while Hyderabad ranks 27th with 151 MW. Mumbai also ranks 11th globally by development pipeline, with nearly 5 GW of capacity in the pipeline, while Hyderabad ranks 19th with more than 2.2 GW.

What Keeps the Cloud Running Efficiently?

A data centre may be a digital asset, but its economics remain stubbornly physical. Servers require electricity. Electricity becomes heat. Heat must be removed to allow servers to work. Power must be quality-conditioned and backed up.

Uninterruptible Power Supply (UPS) batteries must be monitored. Chillers, pumps, cooling towers, Computer Room Air Handler (CRAH) units, switchgear and generators must perform when required, not merely when it is convenient. In an increasingly power-constrained industry, facilities management is becoming one of its principal efficiency levers.

The Data Centre's Most Underestimated Efficiency Lever

Data centres are constructed to exceptional specifications. However, a superior design does not ensure that the facility is operated efficiently. Equipment can deviate from its intended operating parameters. Filters can get blocked. Sensors may lose their calibration. Individually, these appear minor. With thousands of operating hours and tens or hundreds of megawatts, they are anything but.

Power Usage Effectiveness, or PUE, illustrates the point. PUE compares total facility energy consumption with the energy consumed by IT equipment. The closer the number moves towards 1.0, the lesser energy is being consumed by supporting infrastructure. The Uptime Institute's 2026 survey reports an average industry PUE of 1.52, dropping to 1.36 for larger, newer facilities, which tend to be more efficient.

Many modern data centres achieve PUEs of 1.3 or lower. Reducing PUE from 1.5 to 1.4 for a data centre with a steady IT load of 100 MW can save approximately 10 MW in infrastructure power requirements, translating into around 87.6 GWh of electricity savings annually. The cheapest megawatt may be the one you stop wasting.

The Thermal Economics of the Data Centre

Physics of a data centre in Mumbai or Chennai is not the same as of one operating in Scandinavia. High ambient temperatures, humidity, monsoon conditions and local water availability materially influence cooling performance and technology choices.

This is where good FM becomes highly technical. The objective is not simply to "maintain the chiller." It is to understand the entire thermal chain. Operational decisions repeated every hour become financial outcomes:

• Are chilled-water supply and return temperatures being optimised?
• Are variable-speed drives responding properly to actual demand?
• Are Computer Room Air Conditioner (CRAC) and Computer Room Air Handler (CRAH) units competing with one another?
• Is hot-air recirculation occurring because containment has been compromised?
• Are filters creating unnecessary pressure drops?
• Are redundant machines being operated in combinations that make sense at current load?
• Is equipment being rotated intelligently?
• Are environmental set points still appropriate for the installed IT equipment or are teams cooling the room more aggressively simply because that is how it has always been done?

These may sound like operational questions. But repeated every hour, they become financial outcomes.

The Power Equation: Finding Efficiency Before Finding More Power

Much of the current debate around data centres is focused on finding new power. There is another side to that equation: using the power already secured more intelligently. For India, this is particularly important because electricity is not simply another utility input.

Knight Frank's Global Data Centre Atlas 2026 reinforces the point from a different angle, describing the emerging model as “power-first, not location-first” and identifying reliable, scalable power as the key bottleneck to AI infrastructure.

Energy management can no longer mean producing a monthly consumption graph. Operators must understand where electrical losses occur across transformers, UPS systems, power-distribution equipment and cooling infrastructure, whether redundancy configurations are creating avoidable inefficiency, whether asset loading sits within the optimum operating band, and whether deterioration in equipment efficiency can be identified before it becomes visible in the electricity bill.

A serious FM operation should be able to explain not merely how much energy was consumed, but why. That changes the FM mandate. For data centres, facilities management is no longer just about maintaining buildings and equipment; power resilience, uptime, energy efficiency and critical asset performance become central to operational continuity. That is the difference between reporting and managing.

The Hidden Water Footprint of Digital Infrastructure

Energy attracts most of the attention because it arrives with an obvious bill. Water can be easier to ignore. That is unlikely to last. According to the Council on Energy, Environment and Water (CEEW), India’s data centres used approximately 150 billion litres of water in 2024, while both water consumption and electricity use are projected to more than double by 2030.

A typical 100-MW hyperscale data centre can use around 2 million litres of water a day for on-site cooling, depending on design, climate and operating conditions. The significance is greater in India because data-centre growth is clustering around large urban regions already balancing competing demands on water. This means Water Usage Effectiveness, or WUE, should sit alongside PUE on the facilities manager's dashboard.

Cooling-tower cycles of concentration, blowdown optimisation, water chemistry, leakage, treatment-plant performance, rainwater harvesting, recycled-water utilisation and closed-loop systems will increasingly affect both sustainability credentials and operational resilience.

There is no universal answer. Water-cooled systems can improve energy efficiency but increase water dependence. Air-cooled systems reduce water demand but may consume more electricity, particularly in hot climates. Direct-to-chip and other liquid-cooling technologies may change the equation again as rack densities increase.

The task of the FM organisation is therefore not to pursue a single headline metric at all costs. It is to optimise energy, water, resilience and lifecycle cost together. That is a considerably more sophisticated mandate.

The False Trade-Off Between Reliability and Efficiency

There is a persistent belief in critical environments that efficiency must occasionally give way to resilience. Sometimes it must. For an FM organisation, this makes maintenance strategy inseparable from business continuity.

Efficiency is created in the thousands of decisions made between commissioning and decommissioning, which equipment runs and when, how redundancy is configured, when maintenance is performed, how equipment loading is balanced, when an asset is replaced, and how early a developing failure can be identified. These decisions determine not just whether a facility remains operational, but how economically it remains operational.

Predictive models that identify abnormal equipment behaviour before alarms trigger, and maintenance windows scheduled around real redundancy and workload conditions, are where digital facilities management begins to justify the word "digital." The value of technology is not the data itself. It is the decision that the data enables.

The Next Competitive Advantage May Be Operational

India is investing heavily in digital infrastructure. But building the capacity is only the beginning. The next competitive advantage may not come from the data centre that consumes the most power or houses the greatest number of GPUs. It may come from the facility that extracts the most useful computing from every unit of power, water and infrastructure it consumes.

For owners and operators, this means facilities management can no longer be viewed as a function that simply takes over after design and commissioning. It has a direct bearing on operating costs, asset performance, sustainability, resilience, and ultimately the value of the asset itself.

The data-centre boom is here. The FM reckoning must begin now.

Pawan Koyal is Executive Director & Head of Facilities & Asset Management Services at Knight Frank India. Knight Frank FAMS manages over 155 million sq. ft. of commercial and mixed-use assets across India.