Late February a New Jersey City council voted against the construction of an AI data Centre, after protests and concerns from hundreds of residents. The people argued that their electricity and water bills will skyrocket, and the facility would also harm their environment. Now instead of the proposed data centre the site for development will have 600 new apartments, including 10% affordable housing units, warehouses, space for small business, and a park.This is one of the several incidents, where new data centres in the US have been delayed or cancelled. Data Center Watch tracked the opposition to large-scale data centre projects across 28 US states between May 2024 and March 2025. They report that data centre projects worth $64 billion have been blocked or delayed by a growing wave of local, bipartisan opposition. Even MAGA supporters have joined the wave of opposition against the data centres. In its latest polling, Public First found that around 60% of Trump voters are worried about the rapid pace of development of AI. Around 80% of voters think that technology has to be regulated. Farmers in US have refused to give their land for data centres. No wonder the big tech companies spearheading AI in the US are now looking at other countries to expand their data centres. And India has rolled out several red carpets for them. Much before the AI Summit, Budget 2026 promised a tax holiday for 20 years to any country setting up data centres in India. The power and water guzzling nature of data centres, particularly AI data centres is not hidden from anyone. As per an estimate, in 2024 Indian data centres consumed 150 billion litres of water. US data centres, the largest number in the world (about 5,000), guzzled 163.7 billion gallons of water, or 620 billion litres, in 2021 alone. That is equal to the annual water needs of a small town. Companies often underreport their water and power usage, and in several countries there are no mandatory needs of disclosure. In 2023, Google reported that its data centres consume 6.1 billion gallons of water. This need has only grown. Today India has over 270 data centres; Maharashtra leads with 85 data centres, followed by Tamil Nadu at 35 and Karnataka at 32. Of the 32 data centres in Karnataka 31 are in Bangalore, 7 in Whitefield area alone. At this juncture, it is crucial to study the water profile of the tech city, slowly becoming a tanker city due to its acute drinking water crisis. Right now the city is a front-runner in the list of cities to likely run out of drinking water in the near future. According to a study by Deloitte, 1 megawatts (MW) data centre requires 68,500 litres of water every day. A data centre with a capacity of 20 megawatts will need 1.4 million litres of water per day. That much water is consumed in a single day by 27,000 families combined. In 2020, India’s data centre capacity was 520 megawatts. By mid-2025, this has reached 1.5 gigawatts. It is estimated that by 2030, this could reach 4.5 to 6.5 gigawatts. Now, you do the maths.Only 3% of the water on the entire planet is freshwater, and of that, only 0.5% is actually available for people. We also use this for agriculture and other activities. How much of it are we willing to allocate to data centres?Also read: India’s Data Centre Capacity Estimated to Increase Six-Fold, to Put More Pressure on State GridsThe situation is similar in case of power. According to a Financial Express report, the current electricity consumption of India’s data centres is around 10 to 15 terawatt-hours. By 2030, this will be 40-45 terawatt-hours. One terawatt equals 1000 gigawatts, which can light up hundreds of thousands of homes. The share of data centres in the total electricity demand will be around 2.5 to 3% by 2030. This may not seem like a huge percentage, but it is going to impact our already over-loaded state grids. Especially in states like Maharashtra, Tamil Nadu, Karnataka, and Telangana, the peak load on the grid could increase by 5% to 20%. Renewable energy could have helped a bit here, but despite significant investments in renewable energy, the electricity generated there hasn’t reached the grids because the transmission infrastructure is weak.Several reports have now been published on the energy consumption of data centres. They are being compared to carbon emissions. Data centres are no longer just a burden on one area, but have become a major burden on the survival of our entire planet. Data centres in the UK have caused a housing crisis. Permit for new housing is not being released because the grid is at capacity. No wonder you can’t build a house when there is no water and electricity. London Assembly Planning and Regeneration Committee, reported in December 2025 that some new housing developments in west London were temporarily delayed because the electricity grid reached full capacity. In 2024 data centres made up around 10% of UK’s total electricity demand, it is expected to rise by 600% between 2025 and 2050. Water and electricity still get some press, but there are more hidden problems that arise with huge AI data centres. One being the noise factor. The constant hum of servers, network switches, routers, cooling systems can generate sound levels over 80 dBA equivalent to the noise level of city traffic, a hair dryer or a blender. This can cause sleep disturbances, headaches, hypertension, anxiety and even hearing loss, or cardiovascular risks. Another concern is the high volume of e-waste generated by data centres. Servers, batteries and networking equipments requires to be replaced within 3–6 years. The generators and air conditioning system every decade. A study by Nature Computational Science in 2024 estimated that e-waste could reach up to 1.2 to 5.0 million metric tons by 2030, which is 1000 times more than what was produced in 2023. Apart from regular wear and tear, when technology changes, faster chips and better equipments become available, these data-centres would require to scale up to meet more demands of efficiency. Data Centres’ track record on jobsData centres are primarily attractive to the Indian government because their construction immediately creates jobs and visibly kicks up dust. People feel that development is sure to follow. But once built, it has been observed that data centres on an average provide jobs to only 20 to 100 people.Also read: Who Will Bear the Cost of AI Colonialism?In Chile, where companies like Google and Microsoft promised thousands of jobs, since 2012 is a case in point. Rest of World conducted an analysis of 17 data centre projects from 2012 to 2025. These data centres themselves do not disclose how many people they employ. Therefore, based on the environmental review files they filed, it was found that these centres collectively employ only 1,547 full-time employees, and most of the jobs are in security and cleaning.More than data centres, India is seeing a rise in backend jobs of AI, such as data annotation jobs. There are several companies that are employing about 70,000 people mainly in semi-rural areas and town for data annotation jobs. These are the people behind training AI basic things, for example how to differentiate a cat from a tiger, or how to fold a towel. The market value of data annotation industry is estimated at about 250 million dollar. Recently, Guardian reported the dark side of this industry. Young women in small towns are made to watch hours of sexually violent content to train AI to distinguish abusive content from non-abusive one. Without any psychological support and with little pay, and no job security, these women are left with very little options. This is the stark reality of indirect jobs data-centres are going to create in future. Is it possible to steer data centres and AI revolution in an alternate way without putting such high pressure on people and our resources? Some AI experts say yes. Notably, there is also a belief that AI can run on smaller models in an alternate environment. Kavita Kabeer is a writer and a satirist, currently helming the shows ‘Cracknomics’ and ‘Digital Arrest’ for The Wire.