While data centres proliferate across the country to meet the surging demand from the rapid digitisation and the generative AI boom, the reliance on coal-powered electricity threatens to undermine sustainability goals and exacerbate the nation’s water crisis.
Coal remains a dominant source of power generation in India, accounting for over 70% of actual electricity production. Despite significant investments in renewable energy, the country continues to rely heavily on coal to meet its rising power needs.
In the financial year 2022-23, non-renewable sources, primarily coal, generated 75.11% of India’s electricity. This dependence on fossil fuels poses a major challenge for the data centre industry, which requires vast amounts of electricity to operate.
In states hosting the most data centres, the reliance on fossil fuels is even higher.
For example, Maharashtra, which hosts over 40% of India’s data centres, gets 78% of its electricity from fossil fuels. In the Delhi NCR region, 85-90% of electricity comes from fossil fuels.
India’s draft National Data Centre Policy from 2020 encourages but does not mandate using renewable energy for data centres. The policy places the onus of reducing the carbon footprint up to private companies.
The environmental impact of this growing energy consumption is a significant concern that current policies do not sufficiently address. Unless stronger policies are implemented, data centres will continue to rely heavily on coal-generated electricity as they expand.
Microsoft, for example, has committed $3.7 billion to build data centres adding 660 MW of IT capacity in Telangana, equivalent to the annual electricity needs of about half a million European households.
Why Data Centers Need Exorbitant Power
Data centres require significant power to support a variety of workloads and infrastructure components.
The primary power consumers in data centres are the IT equipment running various workloads, such as data processing and storage, cloud computing services, and high-performance computing (HPC) applications like AI/ML, big data analytics, and simulations. Plus, mission-critical enterprise applications (ERP, CRM, databases), and edge computing workloads that require low latency.
Servers, storage systems, and networking gear make up the majority of IT equipment power consumption. Workload density and complexity are increasing, driving up power requirements.
Cooling systems account for a large portion of data centre power usage, often up to 50%. Cooling is needed to combat the heat generated by IT equipment and maintain optimal operating temperatures. The amount of cooling required scales with the IT workload.
Additional power is needed to support data centre infrastructure, such as power distribution and backup systems (UPS, generators), lighting, security and monitoring systems, and fire suppression.
Several factors influence the power consumed by data centre workloads. Servers often run at low utilisation but still consume power even when idle.
To solve these issues, data centres are implementing various strategies to reduce their power consumption.
For instance, they use variable-speed fans that can reduce power consumption by 20%, adopt liquid cooling for high-density racks, and raise operating temperatures to reduce cooling needs (modern IT equipment can run reliably at temperatures up to 77°F/25°C).
Besides, they also implement hot aisle/cold aisle containment, optimise airflow, and use sensors and automation to match cooling capacity to IT workloads in real-time.
Improving the efficiency of IT equipment itself is another key strategy. Servers and storage drives are the largest power consumers. Data centres are adopting newer, more energy-efficient server CPUs and power supplies.
They use solid-state drives (SSDs) that consume about 1/4 the power of hard disk drives, consolidating storage with larger capacity drives, and increasing virtualisation to improve server utilisation, as idle servers still consume significant power.
AI/ML is being leveraged to optimise IT workloads and reduce energy consumption. AI can analyse data centre operations in real-time to identify inefficiencies and provide optimisation recommendations.
Workloads can be shifted in time to align with renewable energy availability or reduce grid demand. Some non-urgent compute tasks can be deferred during times of grid stress as part of demand response programs.
Losses in power distribution and backup systems are being reduced through more efficient uninterruptible power supplies (UPS) and power distribution units (PDUs), as well as by eliminating double-conversions by using line-interactive UPS topologies.
Finally, data centres are increasingly powered by renewable energy to reduce reliance on fossil fuels and carbon emissions.
This includes partnering with utilities on “green grid” initiatives, purchasing renewable energy credits or entering power purchase agreements for solar, wind, etc, and implementing on-site renewable generation like rooftop solar panels.
Giants like NVIDIA and Intel are also contributing to optimise performance for better and more efficient power usage.
Nvidia’s Blackwell GPU architecture, announced in 2024, enables running trillion-parameter AI models at 25 times less cost and energy than the previous Hopper architecture.
A language model previously requiring 8,000 Hopper GPUs and 15 MW can now run on 2,000 Blackwell GPUs using only 4 MW. The GB200 Grace Blackwell Superchip connects two B200 Tensor Core GPUs to the Grace CPU, delivering 1.4 exaflops of AI performance.
Intel used 99% renewable electricity across its worldwide operations in 2023. It completed over 2,000 energy conservation projects in the last decade, saving 4.5 billion kWh.
Intel aims for 100% renewable electricity globally by 2030 and an additional 10% reduction in Scope 1 and 2 carbon emissions from 2020.
Other companies like Google and Cadence are leveraging energy-efficient platforms like Blackwell to reduce energy needs in data centres and computing-intensive workflows.
Also, Drying Up Rivers
Due to geographical reasons data centres in India consume exorbitant water directly for cooling and indirectly through non-renewable electricity generation, exacerbating water scarcity in India, one of the world’s most water-stressed countries.
In Bengaluru alone, data centres use 1.4 crore litres of water daily, equivalent to nearly 41,900 households. Mumbai, Delhi, and Chennai also face an increasing water crisis.
Newer, more efficient IT hardware and cooling systems can reduce power needs. Duplicating components for failover increases total power draw. Planning for growth requires building in extra power capacity.
As traditional air-brd cooling methods have reached limitations, companies are now turning to direct-to-chip liquid cooling for its superior heat transfer capabilities.
It has also prompted the adoption of waterless, two-phase solutions like ZutaCore’s HyperCool, which is being used by AMD, Dell, and data centre giants like Equinix.
HyperCool offers significant benefits for high-performance computing, server densification, and data centre sustainability, tailored to modern cloud, AI, and HPC workloads.
With its dielectric liquid cooling solution, ZutaCore addresses the rising challenges of escalating heat densities exceeding 100 KW in data centres as heavy GPU racks become part of Indian data centres like Yotta.
Vijay Sampathkumar, country manager at Zutacore, explained, “We use zero water… and this technology can ‘decimate 92% of GPU heat’ and ‘70% of CPU heat’ without water.”
Urgent Need for Sustainable Practices
Experts warn that without urgently adopting sustainable practices like waterless cooling technologies, greater renewable energy use, and stricter regulation, the water crisis could force data centres to relocate, jeopardising India’s digital ecosystem.
Some companies are taking concrete steps. Google and Microsoft use recycled wastewater for cooling, CtrlS aims for 100% recycled water, and NTT targets 99% waste recycling by 2030. Innovative cooling solutions like liquid immersion and direct chip cooling are being adopted to slash water usage.
However, much more needs to be done to make India’s data centre growth sustainable. The government must strengthen policies to mandate renewable energy use and water conservation. Companies need to accelerate the shift to waterless cooling technologies and source clean power.
Without urgent action, India’s data centre dream powered by coal could turn into an environmental nightmare. The country cannot afford to sacrifice sustainability for digital ambitions.