AI and Energy Crisis — Data Centre Power Consumption, Grid Stress and Environmental Concerns for UPSC

CURRENT AFFAIRS | MARCH 2026

Introduction

The rapid proliferation of Artificial Intelligence has triggered an unprecedented demand for computational infrastructure, and with it, a correspondingly massive appetite for energy. As nations race to build AI capabilities, a critical question has emerged that receives insufficient attention in policy circles: can the global energy grid sustain the exponential growth of AI without triggering grid instability, accelerating climate change, and exacerbating water scarcity? The intersection of AI and energy policy represents one of the most consequential governance challenges of the coming decade — one that India, with its dual ambitions of AI leadership and sustainable development, must navigate with particular care.

The Scale of AI’s Energy Consumption

Data from the International Energy Agency (IEA) paints a sobering picture. Global data centre electricity consumption has grown at approximately 12 percent per year since 2017 — a rate that far outpaces overall electricity demand growth. The IEA projects that data centre electricity consumption will reach 945 terawatt-hours (TWh) by 2030, a figure that would make data centres one of the largest electricity-consuming sectors globally, comparable to the entire electricity consumption of Japan.

This growth is driven primarily by AI workloads. Training a single large language model can consume as much electricity as a small town uses in a year. The inference phase — where trained models are deployed to serve billions of queries — is even more energy-intensive in aggregate, because it runs continuously at scale. As AI applications proliferate across search engines, autonomous vehicles, medical diagnostics, financial trading, and content generation, the computational (and therefore energy) demands will continue to escalate.

Grid Stress: The Indian Dimension

India’s data centre capacity is expected to reach 2.5 gigawatts (GW) by 2027 — a dramatic expansion from the current base. Major data centre clusters are concentrated in Maharashtra (Mumbai and Pune), Karnataka (Bengaluru), Telangana (Hyderabad), Tamil Nadu (Chennai), and the National Capital Region. This geographical concentration creates localised grid stress: data centres in these regions compete for electricity with residential consumers, industrial units, and commercial establishments.

Indian power grids are already under strain during peak demand periods, particularly in summer months. The addition of large-scale data centre loads — which operate 24/7 and demand extremely reliable power supply — could exacerbate frequency fluctuations, voltage instability, and peak load management challenges. Unlike industrial consumers, which may have flexible load profiles, data centres require near-perfect uptime, necessitating backup diesel generators that add to carbon emissions.

The Water Crisis: AI’s Hidden Resource Demand

Beyond electricity, AI infrastructure has a significant and often overlooked water footprint. Data centres generate enormous amounts of heat, which must be dissipated through cooling systems. Many of these systems rely on evaporative cooling, which consumes vast quantities of water. According to recent estimates, the global AI infrastructure may consume water equivalent to six times Denmark’s annual water use — a staggering figure that has drawn the attention of environmental scientists and water policy experts.

For India — a country where 600 million people face “extreme water stress” according to NITI Aayog — the water demands of data centres pose a direct conflict with agricultural, industrial, and domestic water needs. Data centre operators in water-scarce regions such as Marathwada, Vidarbha, and Rajasthan face both an ethical and a regulatory challenge: how to justify massive water consumption for AI computation in communities that struggle to meet basic drinking water needs.

The Policy Vacuum: India’s Regulatory Gap

India currently lacks a comprehensive national data centre policy. The Draft Data Centre Policy 2020, formulated by the Ministry of Electronics and Information Technology (MeitY), was circulated for public comment but was never formally adopted. This policy vacuum has several consequences:

• No mandatory energy efficiency standards: Unlike the European Union, which has proposed energy efficiency benchmarks for data centres under its Energy Efficiency Directive, India has no compulsory requirements for Power Usage Effectiveness (PUE) ratios, renewable energy sourcing, or waste heat recovery.
• Fragmented state-level policies: Several states — including Maharashtra, Telangana, Karnataka, Tamil Nadu, and Uttar Pradesh — have introduced their own data centre policies, but these focus primarily on investment incentives (land subsidies, tax breaks, expedited clearances) rather than environmental safeguards. The result is a race to the bottom, where states compete to attract data centre investments without imposing meaningful sustainability requirements.
• Institutional capacity gaps: State electricity utilities often lack the technical expertise to model the impact of data centre loads on grid stability, negotiate power purchase agreements with data centre operators, or implement smart grid technologies that could optimise load distribution.

International Comparisons: How Other Nations Are Responding

The European Union has taken the most proactive approach, requiring data centres above a certain capacity to report their energy consumption, water usage, and waste heat recovery rates. The EU’s proposed Energy Efficiency Directive includes specific provisions for data centre operators. Singapore imposed a temporary moratorium on new data centre construction in 2019 to assess environmental impacts, lifting it only after implementing a green data centre framework. Ireland, where data centres already consume approximately 18 percent of national electricity, has introduced planning restrictions in certain grid-constrained areas.

India can draw valuable lessons from these international experiences while tailoring its approach to domestic circumstances — including the need for rapid digital infrastructure growth, the constraints of the power grid, and the imperative of water conservation.

Way Forward: An Integrated AI-Energy Policy Framework

Several policy interventions merit consideration:

• Designate data centres as strategic energy consumers: Just as the Electricity Act, 2003, provides special frameworks for railway traction loads and defence establishments, data centres — given their critical role in national AI infrastructure — should be classified as strategic consumers with dedicated power supply arrangements, demand-response obligations, and renewable energy purchase requirements.
• Nuclear power and Small Modular Reactors (SMRs): The baseload, 24/7 power requirements of data centres align well with nuclear energy’s generation profile. Small Modular Reactors, which are compact, factory-built, and can be deployed near demand centres, offer a promising pathway for powering data centres with low-carbon electricity. India’s Department of Atomic Energy is exploring SMR designs, and aligning this programme with data centre energy needs could accelerate both nuclear deployment and sustainable AI development.
• Treated wastewater for cooling: Data centres should be mandated (or incentivised) to use treated wastewater rather than freshwater for cooling purposes. This approach simultaneously reduces pressure on freshwater resources and creates a productive use for treated wastewater, which is currently underutilised in most Indian cities.
• Mandatory energy efficiency standards: The Bureau of Energy Efficiency (BEE) should develop and enforce PUE standards for data centres, with escalating stringency over time. Data centres that exceed benchmark PUE ratios should face financial penalties or be required to purchase renewable energy certificates.
• Grid modernisation: State electricity utilities must invest in smart grid technologies that can dynamically manage data centre loads, integrate renewable energy sources, and provide demand-response capabilities. This requires both capital investment and human capacity-building.

The Broader Ethical Question

As OpenAI CEO Sam Altman observed, “Governments should focus on regulating potentially catastrophic AI issues and be lenient on lesser important issues.” While this perspective has merit in the context of AI safety, the energy and environmental implications of AI are not “lesser” issues — they are fundamental to the sustainability of the entire AI enterprise. A global AI industry that accelerates climate change and depletes water resources undermines the very human welfare it claims to advance.

Conclusion

The energy crisis triggered by AI’s computational demands is not a future hypothetical — it is a present reality. India, which aspires to be both an AI leader and a climate-responsible nation, faces a particularly acute version of this challenge. Addressing it requires an integrated policy framework that treats AI infrastructure, energy planning, water management, and environmental protection as interconnected dimensions of a single governance challenge. For UPSC aspirants, the AI-energy nexus is an emerging theme that bridges science and technology, environment, economy, and governance — precisely the kind of cross-cutting analysis that distinguishes exceptional Mains answers.

Source: UPSC Essentials, The Indian Express — March 2026. Content rewritten and analysed for UPSC preparation by Civils Gyani — Empowering Future Officers.

Practice Quiz

Test your understanding with these 10 MCQs:

Quiz data error: Control character error, possibly incorrectly encoded