Growing Interconnections Between Energy and AI Worldwide

The Energy-AI Nexus: Driving Progress or Consuming Resources?

945 TWh — that’s the electricity demand projected for data centres worldwide by 2030, more than double the consumption in 2020. Roughly 40% of that growth will be driven by AI-optimised centres alone. The International Energy Agency (IEA), in its 2024 report, flagged this staggering increase, illustrating AI's energy appetite amid ambitious global transitions to renewables. But behind these numbers lies a dilemma: is the synergy between energy and AI a sustainable partnership or an ecological gamble?

The Policy Instrument: India’s Strained Energy Landscape

India, the world’s third-largest energy consumer, stands at the crossroads of AI and energy convergence. According to McKinsey, Indian data centres, currently demanding 1.2 GW of capacity, will require 4.5 GW by the end of this decade. AI-driven operations alone could consume 40–50 TWh annually by 2030. Mumbai will account for 41% of this capacity, followed by Chennai (23%) and the NCR (14%). Despite recent strides in renewables, India’s energy mix remains dominated by coal and natural gas—raising critical questions about the sustainability of scaling AI-enabled infrastructures.

This consumption demand comes alongside water stress and escalating e-waste concerns. Cooling servers now necessitates millions of litres of water daily, aggravating regional resource imbalances. The turnover of AI-specific GPUs and TPUs accelerates electronic waste, testing India's underdeveloped waste management systems. While institutional mechanisms like the Energy Conservation Building Code aim to integrate AI towards efficiency, implementation gaps persist.

The Case For: AI as a Catalyst for Sustainable Energy

Despite its mammoth energy footprint, AI promises significant environmental dividends. Take smart grids: AI enables predictive load balancing, enhancing the efficiency of batteries and renewable integration. For instance, AI-driven forecasting models have reduced curtailment rates in wind and solar plants by up to 20% globally, according to IEA data. In climate modelling, AI assists adaptation strategies, enabling improved disaster prediction and infrastructure planning. India’s National Energy Efficiency Roadmap already incorporates AI to streamline energy usage across real estate and manufacturing sectors, with reforms like automated HVAC systems achieving up to 25% energy savings.

AI’s systems-level contributions extend to transportation and logistics. Routing algorithms reduce fuel use and emissions, while adaptive supply chains align distribution patterns with demand spikes. Globally, this suggests a tangible reduction in carbon footprints, at least in sectors where such tools are actively deployed. The merits of AI, when paired with clean energy sources, are undeniable.

The Case Against: Unsustainable Resource Costs

The irony here is AI’s dual character — as both solution and stressor. While aiding renewable integration, global AI adoption risks outpacing these gains. Training large AI models like ChatGPT requires data centre operations that demand exponential energy inputs concentrated on fossil-fuel-heavy grids. If renewables fail to scale at comparable speeds, CO₂ emissions could balloon. As a dangerous parallel, cooling technologies overburden freshwater resources, jeopardizing hydrological balances in water-stressed regions like Maharashtra.

Institutionally, India faces structural obstacles. Implementation of AI-driven efficiencies is overwhelmingly skewed towards urbanized hubs, failing to extend equitable benefits to smaller cities and rural areas. Meanwhile, regulatory gaps in e-waste policies, combined with India’s reliance on imported GPUs, leave critical vulnerabilities unaddressed. Without a comprehensive regional sustainability strategy, AI could deepen existing resource divides rather than mitigate them.

Learning from International Experiences: Sweden's Energy-AI Model

Sweden offers a pointed comparison here. As part of its AI adoption strategy, Sweden prioritized data centre sustainability by coupling operations exclusively with renewable energy sources, including its abundant hydropower. Result? Sweden has managed to keep data centre energy costs below EU averages while achieving nearly net-zero emissions from AI-related electricity use. This contrasts sharply with India’s reliance on coal-heavy grids, which amplify ecological risks.

Still, even Sweden’s model isn’t flawless. Water usage for server cooling remains problematic, drawing from freshwater reserves that strain regions during dry periods. Furthermore, lower global GPU turnover has kept its e-waste generation lower than in India, but this balance is precarious as AI models grow more computationally intensive.

Where Things Stand

This debate boils down to pace versus sustainability. India’s AI ambitions—highlighted through its digitalisation and smart city drives—are not inherently at odds with sustainability goals. However, the concentration of data centre demand near water-stressed areas (Mumbai, Chennai), combined with slow renewable integration, reveals troubling structural limitations. The challenge is multi-tiered: policy must align AI-growth with energy transition timelines while also tackling distributional inequities. Immediate corrections in regulatory frameworks for e-waste and water dependencies should be non-negotiable.

AI can drive environmental and economic benefits if framed responsibly within the larger climate agenda. It is not merely a technological force but a governance challenge — a potential accelerant or a liability depending on choices made today.

Practice Questions for UPSC

Prelims Practice Questions

Frequently Asked Questions