Artificial Intelligence (AI) has become the most convenient tool to satisfy our curiosity, answer questions, and bring creativity to life. It is transforming the global economy, political systems, social relations, technological use, and the way knowledge is created and shared.
Today, from learning small facts to conducting deep research, we increasingly rely on AI tools such as ChatGPT. Chat-based AI platforms are clear evidence of this shift—ChatGPT alone answers billions of queries every day. However, as AI grows in development and usage, its challenges and risks are becoming more evident.
AI is reportedly reducing human intellectual and logical abilities and raising concerns about potential threats to humanity itself—even its creators have started issuing warnings. Another growing concern is AI’s impact on the environment, particularly its heavy exploitation of one of the planet’s most essential natural resources—water. Discussions about the environmental downsides of AI technology are now gaining worldwide attention.
AI consumes vast amounts of energy, which increases carbon emissions. Data centers require enormous quantities of water for cooling, while manufacturing advanced AI technologies and equipment generates electronic waste and depletes natural resources. This combination of energy use, water demand, and resource consumption poses serious environmental risks.
Table of Content
- Water Scarcity and AI’s Growing Demand
- Data on Water Consumption
- The Issue of Non-Recyclable Water
- Nepal’s Efforts
- Other Countries’ Impacts and Responses
- Conclusion
Water Scarcity and AI’s Growing Demand
Half of the world’s population is already experiencing water shortages. Climate change, population growth, and technological expansion are intensifying this crisis. AI platforms and chatbots rely on massive, sophisticated data centers with thousands of computer servers to process user queries and generate AI content.
These data centers consume huge amounts of energy, which leads to heat buildup that must be cooled with significant water usage. AI chatbots consume more electricity than traditional search engines, producing more heat and therefore requiring even more cooling. Water is used not only for cooling data centers but also in electricity production for running AI servers.
A study reveals that producing just one microchip used for AI consumes about 2,200 gallons of water.
Data on Water Consumption
According to Sam Altman, CEO of OpenAI, ChatGPT uses one teaspoon of water for every 15 questions it answers. A study by American researchers shows that newer, more advanced ChatGPT versions use between 2 and 10 teaspoons of water per query.
Training models such as GPT-3 requires millions of liters of water. The International Energy Agency (IEA) estimates that answering one AI query consumes roughly 10 times more energy than a Google search, which further increases water demand.
OpenAI reports that ChatGPT answers over one billion queries daily. With dozens of other AI platforms operating globally, the total water consumption reaches alarming levels.
-
Current global data centers consume 560 billion liters of water annually
-
By 2027, data center water usage is projected to reach 6.6 trillion liters
-
AI-specific data centers could consume 1.2 trillion liters of water annually by 2030
-
AI industry water use is expected to increase sixfold over the coming years
The United Nations reports that the number of data centers worldwide has grown from 500,000 in 2012 to over 8 million today. IEA projects that by 2030, water consumption by all data centers will nearly double. In the United States, 12% of all electricity will be used by data centers by 2026.
A single 100-megawatt data center consumes more electricity than 75,000 households and uses 2 million liters of water daily. Google reported that its data centers consumed 3.7 billion liters of water in 2024, of which 2.9 billion liters evaporated. This amount of water could provide 50 liters per day for 1.5 million people for a year.
The Issue of Non-Recyclable Water
Water used for cooling in data centers is largely non-reusable. Around 80% evaporates, leaving only 20% available for reuse. This makes AI technology a major contributor to water depletion and pollution.
While technology has made our lives easier and AI is achieving remarkable feats, it also brings significant negative environmental consequences.
AI can still play a crucial role in climate research, energy efficiency, and biodiversity conservation. However, it must be integrated with global sustainability efforts. This includes developing technologies to reduce water use in data centers, adopting renewable energy, reusing equipment, and minimizing carbon and water footprints.
Responsibility lies not only with big tech companies but also with everyday internet users, who can help by reducing unnecessary and excessive AI usage to conserve energy and water.
Nepal’s Efforts
Nepal has one of the fastest rates of adopting new technologies. The country is already earning billions of rupees annually through IT exports and is embracing AI to build a human-centered, ethical, and prosperous future. The government has introduced the AI Policy 2082 with a vision to accelerate economic and social development through safe and efficient use of AI technology.
However, the policy does not yet fully address the long-term environmental consequences of AI. It prioritizes building advanced AI infrastructure, such as data centers, cloud systems, and AI hardware industries.
The policy does mention establishing data centers in Nepal’s high-altitude and Himalayan regions using green infrastructure, showing some consideration for reducing environmental risks. It also promotes using AI for environmental protection, water and weather forecasting, disaster management, and pollution control—signs of a long-term vision to link AI with sustainability.
Other Countries’ Impacts and Responses
United States
AI development hubs rely on large cloud campuses that drive up water and power demand. Microsoft disclosed a 34% year-over-year jump in global water use from 2021 to 2022—researchers linked the spike to AI training activity concentrated in Iowa, where an Azure supercomputer trains frontier models.
Several U.S. data center sites also report high cooling water withdrawals. Public tallies show Google facilities consuming hundreds of millions of gallons annually across states such as South Carolina, Oklahoma, Oregon, Georgia, and Tennessee.
Ireland
Ireland’s rapid data center build-out has shifted national electricity demand. Official statistics show data centers rising from 5% of total metered electricity in 2015 to 21% in 2023, overtaking all urban homes; newer estimates place 2024 usage at roughly 22%. Grid planners now forecast strong demand growth and have imposed location and connection constraints around Dublin.
Policy debate centers on balancing AI-driven investment with security of supply and climate targets; approvals in Dublin are paused until grid upgrades and renewable additions catch up.
Singapore
After a pause on new data centers, Singapore relaunched growth under strict sustainability rules. The Green Data Centre Roadmap targets at least 300 MW of new capacity tied to high efficiency and lower water use, with additional green-energy deployments planned.
Operators must adopt energy- and water-efficient cooling to secure permits in the land- and resource-constrained city-state.
Netherlands
Local reporting highlights high water withdrawals at hyperscale sites. In Eemshaven, a Google facility used about 232 million gallons in 2023, adding pressure on regional water planning amid growing AI loads.
Google publishes efficiency metrics for several European sites as part of its transparency efforts.
United Kingdom
Recent UK analyses warn that data centers’ water use is significant and rising, with implications for drought-prone catchments and regional planning.
Government and industry briefings urge clearer reporting, siting guidance, and technology choices that minimize potable water use and prioritize recycled sources.
Global Outlook
International Energy Agency assessments and national briefings converge on a trend: AI-heavy data centers will materially increase electricity demand this decade, with knock-on effects for water and emissions.
Planning responses include renewable power procurement, high-efficiency cooling, heat reuse, and siting in cooler climates or near non-potable water sources.
Conclusion
AI offers immense benefits, but its energy and water demands create serious environmental risks. Sustainable solutions—like renewable energy, efficient cooling, and responsible usage—are essential to balance AI progress with protecting natural resources.
Environment and Ecology Artificial intelligence (AI)