How can India measure and mitigate the environmental impact of AI?

India can adopt a multi-pronged approach to measure and mitigate AI’s environmental impact. Measurement Standards: Establish uniform standards to track energy consumption, GHG emissions, water usage, and resource utilization of AI systems, involving stakeholders such as tech companies, think tanks, and NGOs. Data Collection: Deploy sustainability metrics for AI, including lifecycle assessments of models, server energy efficiency, and freshwater usage, similar to OECD and UNEP studies. Policy Integration: Incorporate AI environmental assessments into EIA procedures and ESG disclosure norms, drawing inspiration from the EU’s CSRD framework that mandates emissions reporting for high-compute activities. Adoption of Sustainable Practices: Use pre-trained models, power data centres with renewable energy, optimize algorithms for efficiency, and promote responsible AI usage to reduce the carbon footprint. Such a structured approach ensures accountability, facilitates global alignment, and promotes AI deployment in a sustainable manner.

The Hidden Carbon of Artificial Intelligence: Unseen Environmental Costs of AI

Q: What are the environmental impacts of developing and deploying Artificial Intelligence (AI) models?

The environmental impacts of AI primarily stem from the high energy and resource requirements for training, deploying, and maintaining AI systems. Large AI models, especially Large Language Models (LLMs) , consume significant electricity, which contributes to greenhouse gas (GHG) emissions. For example, a single LLM can generate nearly 300,000 kilograms of CO 2 , equivalent to emissions from five cars over their lifespan. AI data centres also have a substantial water footprint . UNEP estimates that AI servers may use 4.2 to 6.6 billion cubic meters of water by 2027, potentially worsening water scarcity. Additionally, even day-to-day use of AI tools such as ChatGPT consumes 10 times more energy per request than a typical Google search. These factors collectively contribute to climate change and resource depletion, highlighting the need for sustainable AI practices.

Q: What factors contribute to the high energy and carbon footprint of AI models?

Several factors contribute to AI’s environmental impact. 1. Model Complexity: Training large AI models requires extensive computational resources, often involving thousands of GPUs over weeks or months, leading to high electricity consumption. 2. Data Intensity: AI relies on massive datasets, which must be stored, processed, and transmitted, further increasing energy demand. 3. Server Infrastructure: Data centres supporting AI require continuous power for operation and cooling, resulting in significant energy use and water consumption. For example, studies have shown that training a single NLP model emits over 626,000 pounds of CO 2 . While individual AI prompts (like a single ChatGPT query) may consume low energy, the cumulative effect of millions of queries globally magnifies the carbon footprint. These factors illustrate the need for energy-efficient AI design and responsible deployment.

From energy hungry servers to water scarcity, AI’s rise brings climate challenges India and the world can no longer ignore

Surya

•January 14, 2026•4 mins read

“AI environmental impact, carbon footprint, energy, water, sustainability, India, global policies.”

Not Started

1. Context and Emergence of the Issue

Artificial Intelligence (AI) is rapidly transforming sectors ranging from healthcare to agriculture, enhancing efficiency and innovation. However, its environmental implications—particularly energy consumption, water use, and greenhouse gas emissions—remain underexplored. The OECD working paper, “Measuring the Environmental Impacts of Artificial Intelligence Compute and Applications”, highlights that AI algorithm development comes with substantial carbon footprints, contributing to climate change challenges.

The global ICT industry, which encompasses AI-related computing, is estimated to account for 1.8%–2.8% of global greenhouse gas emissions, with some studies indicating figures as high as 2.1%–3.9%. While corporate reports, such as Google’s claim that a single AI prompt consumes only 0.24 watt-hours, suggest minimal environmental impact, these figures are often contested due to incomplete assessments.

The issue is gaining attention as the environmental costs of AI scale with the deployment of Large Language Models (LLMs) and high-compute algorithms. AI’s energy-intensive processes, if unchecked, can exacerbate climate risks and resource scarcity, particularly in water-stressed regions.

Recognizing AI’s environmental footprint is essential for governance; ignoring it may undermine sustainability targets and lead to unanticipated climate and resource stresses.

2. Environmental Impacts of AI

The UNEP (2024) issue note emphasizes that AI servers could consume 4.2–6.6 billion cubic meters of water by 2027, raising concerns about freshwater scarcity. Training a single LLM can generate ~300,000 kilograms of carbon emissions, while other studies estimate 626,000 pounds of CO₂ per model, equivalent to emissions from five cars over their lifespan.

The operational use of AI, such as virtual assistants like ChatGPT, also entails significant energy use. A UNEP study indicates that an AI request consumes 10 times more energy than a standard Google search, illustrating the cumulative environmental burden of AI deployment.

Impacts:

High energy consumption contributes to increased greenhouse gas emissions.
Water usage for cooling servers could intensify local water stress.
Carbon emissions from training large AI models add to global climate change pressures.

Understanding these environmental impacts is critical for sustainable policy; failing to do so could result in unchecked carbon emissions and resource depletion.

3. Global and National Policy Responses

Internationally, AI governance frameworks are beginning to integrate environmental considerations. The UNESCO Recommendation on the Ethics of Artificial Intelligence (2021) highlights the need to recognize negative environmental impacts. The US Artificial Intelligence Environmental Impacts Act (2024) and the EU harmonized AI rules propose legislative approaches to mitigate AI’s ecological footprint.

In India, discussions around AI and climate change largely focus on leveraging AI for environmental protection rather than addressing its development-related demerits. Extending existing frameworks such as the EIA Notification, 2006, and integrating AI-specific ESG disclosure standards could ensure that India accounts for AI’s carbon and resource costs.

Policy measures:

Adoption of standards to measure AI’s environmental impact involving tech companies, think tanks, and NGOs.
Collection of sustainability metrics including GHG emissions, energy, and water usage.
Inclusion of AI in ESG disclosure frameworks, drawing from the EU’s CSRD model.

Policy interventions are necessary to quantify and mitigate AI’s ecological costs; ignoring this could compromise India’s climate commitments and sustainable development goals.

4. Institutional and Stakeholder Roles

Effective governance of AI’s environmental impact requires collaboration among multiple institutions and stakeholders:

UNEP and OECD provide global research and standards for AI environmental impact measurement.
Indian regulatory bodies, including the Ministry of Corporate Affairs and SEBI, can mandate ESG disclosures for AI-related emissions.
Tech companies developing large-scale AI, alongside think tanks and NGOs, play a crucial role in standard-setting, monitoring, and mitigation practices.

Comparative examples:

EU mandates emissions reporting for data centers and high-compute AI tasks through CSRD.
India can replicate such frameworks to ensure transparency and informed policy decisions.

Institutional coordination is key to sustainable AI governance; lack of stakeholder engagement may result in inconsistent standards and ineffective mitigation.

5. Mitigation and Sustainable Practices

Several strategies exist to minimize AI’s environmental footprint while maintaining its developmental benefits:

Deploy pre-trained models to reduce redundant computational training.
Use renewable energy to power AI data centers.
Report AI-specific emissions metrics to enable informed policy and corporate decisions.

Impacts:

Reduction in energy and carbon emissions from AI operations.
Conservation of water resources used in server cooling.
Promotion of sustainable AI innovation aligned with global climate goals.

Adopting sustainable practices ensures AI contributes to economic growth without compromising environmental objectives.

6. Conclusion

AI’s environmental footprint is a critical policy challenge for India and the world. Effective governance requires measurement, regulation, and sustainable practices, integrating ESG disclosures and international best practices. By addressing AI’s ecological costs proactively, India can harness AI for innovation while advancing climate resilience, resource sustainability, and responsible technological development.

Quick Q&A

Everything you need to know

The environmental impacts of AI primarily stem from the high energy and resource requirements for training, deploying, and maintaining AI systems. Large AI models, especially Large Language Models (LLMs), consume significant electricity, which contributes to greenhouse gas (GHG) emissions. For example, a single LLM can generate nearly 300,000 kilograms of CO₂, equivalent to emissions from five cars over their lifespan.

AI data centres also have a substantial water footprint. UNEP estimates that AI servers may use 4.2 to 6.6 billion cubic meters of water by 2027, potentially worsening water scarcity. Additionally, even day-to-day use of AI tools such as ChatGPT consumes 10 times more energy per request than a typical Google search. These factors collectively contribute to climate change and resource depletion, highlighting the need for sustainable AI practices.

India, as a rapidly growing technology hub, is increasingly developing and deploying AI solutions across sectors like healthcare, agriculture, and governance. Ignoring the environmental costs of AI could exacerbate climate change challenges, resource scarcity, and energy demand, undermining national sustainability goals.

Moreover, India has a legal framework for Environmental Impact Assessments (EIA), but its scope has traditionally focused on physical projects such as infrastructure or industrial facilities. Extending EIAs to AI development would ensure accountability, enable measurement of carbon, water, and energy footprints, and align India with international standards.

Recognizing AI’s environmental impact also supports informed policymaking, incentivizes sustainable practices among tech companies, and positions India as a responsible player in global AI governance frameworks.

India can adopt a multi-pronged approach to measure and mitigate AI’s environmental impact.

Measurement Standards: Establish uniform standards to track energy consumption, GHG emissions, water usage, and resource utilization of AI systems, involving stakeholders such as tech companies, think tanks, and NGOs.
Data Collection: Deploy sustainability metrics for AI, including lifecycle assessments of models, server energy efficiency, and freshwater usage, similar to OECD and UNEP studies.
Policy Integration: Incorporate AI environmental assessments into EIA procedures and ESG disclosure norms, drawing inspiration from the EU’s CSRD framework that mandates emissions reporting for high-compute activities.
Adoption of Sustainable Practices: Use pre-trained models, power data centres with renewable energy, optimize algorithms for efficiency, and promote responsible AI usage to reduce the carbon footprint.

Such a structured approach ensures accountability, facilitates global alignment, and promotes AI deployment in a sustainable manner.

Several factors contribute to AI’s environmental impact.
1. Model Complexity: Training large AI models requires extensive computational resources, often involving thousands of GPUs over weeks or months, leading to high electricity consumption.
2. Data Intensity: AI relies on massive datasets, which must be stored, processed, and transmitted, further increasing energy demand.
3. Server Infrastructure: Data centres supporting AI require continuous power for operation and cooling, resulting in significant energy use and water consumption.

For example, studies have shown that training a single NLP model emits over 626,000 pounds of CO₂. While individual AI prompts (like a single ChatGPT query) may consume low energy, the cumulative effect of millions of queries globally magnifies the carbon footprint. These factors illustrate the need for energy-efficient AI design and responsible deployment.

Several countries and international organisations are actively addressing AI’s environmental footprint. For instance, the United States proposed the Artificial Intelligence Environmental Impacts Act of 2024, while the European Union passed a resolution on harmonized AI rules, integrating sustainability considerations.

UNESCO’s 2021 recommendation on the ethics of AI emphasized recognizing negative environmental impacts as part of ethical AI development. Additionally, OECD and UNEP have published guidelines and working papers measuring the lifecycle emissions and resource consumption of AI, highlighting practices to reduce carbon and water footprints.

These examples demonstrate a growing global consensus on sustainable AI, which India can adapt to its policy frameworks to ensure both technological advancement and environmental responsibility.

India faces multiple challenges in regulating AI’s environmental impact.
1. Measurement Difficulties: Accurate measurement of AI-related emissions, energy consumption, and water usage is complex due to the distributed nature of computing and lack of standardized reporting methodologies.
2. Industry Compliance: Tech companies may resist additional reporting and operational constraints, fearing increased costs or competitive disadvantage.
3. Policy Integration: Extending the Environmental Impact Assessment (EIA) framework to intangible AI systems requires significant adaptation and expertise, as current EIA norms focus on physical projects.

Overcoming these challenges necessitates collaboration between government, industry, and civil society, adoption of international best practices, development of standardized metrics, and incentivization for sustainable AI practices to balance innovation with environmental sustainability.

ESG (Environmental, Social, Governance) disclosure standards offer a mechanism for transparency and accountability in AI deployment. By mandating the reporting of AI-related emissions, energy consumption, and water usage, the Ministry of Corporate Affairs and SEBI can encourage companies to adopt sustainable AI practices.

For instance, the EU’s Corporate Sustainability Reporting Directive (CSRD) framework requires data centres and high-compute activities to disclose emissions. Indian companies could similarly report energy and water use for AI model training and operation, enabling stakeholders to assess environmental impacts and incentivize efficient practices.

This approach promotes responsible AI deployment by linking environmental performance to corporate governance, encouraging the adoption of renewable energy, pre-trained models, and optimization techniques. Over time, ESG integration can make sustainable AI development a standard business practice, reducing India’s carbon and water footprint in the technology sector.

Attribution

Original content sources and authors

Amar Patnaik

The Hindu

Comments (0)

Please sign in to comment

No comments yet. Be the first to comment!