STUDENT OPINION
The Invisible River Behind Your Prompt

Parekar Yohann Vivek - MPP 2028
Published on : Nov 13, 2026
As Google and the Adani Group plan to execute a $15 billion data centre in Visakhapatnam, local banners bearing words like “We cannot drink data” have sparked a debate on how thirsty a data centre is and the rationale behind setting up those centres in water-stressed regions.
This raises an important question: in our pursuit of artificial intelligence, are we creating progress or a form of artificial humanity that overlooks its wider consequences? The problem is not data centres, it is the uncontrollable rise of hyperscale data centres in water-stressed regions of the country. We need data centres as a catalyst to the economic development of our country. Around 65% of Indians use AI, more than double the global average. So, yes we need intelligence but not at the cost of sustainability and humanity.
The ubiquitous statement, “Your Google search, your AI prompt, your doomscrolling is costing water” is misplaced. This accounts for only 3-4 % of the water footprint. For example, if India has 1000 BCM of water available for use, then 80-90% of it goes to agriculture. So, our individual footprints were never really the issue. The issue is the rampant set up of hyperscale data centres without due consideration being given to natural habitats and the water basins that are already running dry. The recent project by Adani and Google of setting up an AI campus in Vishakhapatnam raises eyebrows. Around 480 acres of land has been allocated for this project. The government in February 2026 proposed a tax holiday for foreign companies that route global cloud services through data centres based in India and that too till 2047 – A 21 year tax break! While this may benefit big tech companies, what does it mean for the city and its people?
Beyond Your Prompt: Understanding the Water Footprint of AI
A data centre’s water footprint is calculated by the sum of three categories - “on-site water usage, water use by power plant facilities that supply power to data centres, and water consumption during the manufacturing process of processor chips”. Just to clarify, the “water consumption” in the context of data centres refers to the water withdrawn from our sources minus the water discharged as leftover wastewater after the cooling. The water consumed is the water that is evaporated or immediately taken out of human usage.
Researchers at The Green Grid, a nonprofit industry consortium, developed a metric called Water Usage Effectiveness (WUE) to measure efficiency of water usage by data centres. WUE is reported in liters per kilowatt-hour (kWh): a data centre’s total water consumption, measured in liters, is divided by the total energy consumed by that data centre in kilowatt-hours in the same time period. While “0” is the ideal WUE score, this can only be achieved in air-cooled data centres, and most data centres cannot meet this target due to their location’s climate conditions.
Large data centres can consume up to 5 million gallons of water per day equivalent to water usage by a town of 10,000 - 50,000 people. The Niti Ayog reports that nearly 600 million people in India face high to extreme water stress. Indian data centres consumed around 150 billion litres of water last year and that number is expected to grow by 358 billion by 2030. Approximately 80% of the water (typically freshwater) used by the data centres is evaporated and the rest is discharged as municipal wastewater.
The Cost of Progress: Land, Livelihoods and Local Communities
Recently, CM Naidu passed a remark, and I quote “People unaware of the data centre’s potential are criticising Google’s project.” There’s no denying that the project is important to the city with respect to economic development, but what about the land that is being allocated to this. Reports say the land belongs to the dalits - 266.60 acres at Tarluvada, 160 acres at Adavivaram & Mudasarlova, and 174.80 acres at Rambilli following a request from Raiden Infotech India Pvt. Ltd., a subsidiary of Google. A significant portion of these lands belong to the dalits since the government allocation in the 1970s. Local dalit farmers have been actively protesting out of fear of losing their livelihoods and facing displacement.
“We had to clear the jungle ourselves; it took us 15 years, 15 long years. We worked hard, shed our blood for this land. We built it with pain and determination,” recalls Relli Ramakrishnan, speaking to the Environmental Reporting Collective (ERC).
Cooling the AI Race: Can Technology Offer a Solution?
Coming back to the undisputed usage of water to cool these data centres, there have been multiple news headlines of water shortages in our country, So the argument that there is plenty of water available doesn’t stand. Water is used to cool these data centres while communities struggle to access water for everyday needs. Most of the centres have the open loop cooling system which drains away a lot of water turning it into wastewaters. The alternative to this could be closed loop cooling systems which recycle the water and reuse it for cooling. But the caveat is that this is expensive and the AI race doesn’t permit companies & governments to invest in this system as it takes more time to set up than the open loop cooling system.
There’s no denying that cooling data centres is a complex operation. In areas with limited water availability, server cooling is the best option as it directly cools the GPU and CPUs whereas the air cooling system is more water intensive. Immersive cooling can be practiced involving bathing data centres, servers in a specialized liquid (non-conductive liquid) which absorbs the heat of the chips and then through the help of a heat exchanger, the cool water can be directed back to the tank. There’s also another avenue to be explored: China using oceans to cool their data centres. Can peninsular India with its cool oceans benefit from it? A study conducted by IIT Bombay explores a solution of harnessing the naturally cold temperatures of the deep ocean. Deep seawater cooling systems could alleviate strain on the electrical grid, lower operational costs for business and contribute substantially in reducing greenhouse gas emissions.
Again, the problem is not data centres; they are useful, important for economic development, but if built correctly. The problem is we are greenlighting data centres when they are not optimised for efficiency, nor suitable to the environment and choosing areas where the water basin is already running dry. We need to bring in mandates that regulate these data centres, the need of the hour is to look beyond the draft policy of 2025. Better cooling solutions can reduce environmental impacts, but what is equally important is the concern of land acquisition, displacement and the suitability of the location.
India can fix this. Every problem has a solution that has been implemented elsewhere. The need of the hour is to look at both technological and governance solutions. If we can look outside and adapt those solutions, this can be solved. Because creating intelligence at the cost of humanity is promoting artificial humanity which doesn’t take into account the ripple effects of the rampant evolution and spread of AI.
*The Kautilya School of Public Policy (KSPP) takes no institutional positions. The views and opinions expressed in this article are solely those of the author(s) and do not reflect the views or positions of KSPP.
Rudraram, Patancheru Mandal
Hyderabad, Telangana 502329
