Artificial intelligence may live in the cloud, but its footprint is firmly on the ground. Every AI query and training run depends on data centres filled with power-hungry chips and cooling systems. The debate is no longer only about what AI can do. It is about who supplies the electricity, water and land, and who pays when infrastructure must expand.
The International Energy Agency says data-centre electricity use reached about 485 TWh in 2025 and could rise to roughly 950 TWh by 2030. Electricity consumption from AI-focused data centres could triple. AI’s benefits may travel globally, but many physical costs arrive first on one grid, one water system and one neighbourhood.
Key Highlights
- Global data-centre electricity use could roughly double between 2025 and 2030.
- US data centres directly consumed about 66 billion litres of water in 2023.
- Protests in India, the US and Europe are pushing resource use into politics.
- Developers should fund infrastructure created specifically for their projects.
AI Is Becoming an Energy Story, Not Just a Technology Story
The IEA’s 2026 update says data-centre electricity demand jumped 17% in 2025. A US Department of Energy-backed 2025 update estimates data centres could account for 11.8% of American electricity use by 2030.
A privately owned data centre can still require new transmission, substations and transformers. If those upgrades enter general utility costs, households and smaller businesses may help finance infrastructure created for a few large customers.
A stronger model would require:
- large-load tariffs reflecting grid-upgrade costs;
- long-term power commitments before final approval;
- public disclosure of expected peak electricity demand.
AI growth need not stop. Its infrastructure bill should not appear quietly on somebody else’s monthly statement.
Water Is Becoming the Harder Local Argument
Cooling high-performance servers can consume large volumes of water depending on climate and system design. Lawrence Berkeley National Laboratory’s research data found US data centres used about 66 billion litres directly in 2023, with hyperscale and colocation facilities responsible for 84% of the total.
India shows why national totals can hide local pressure. Reuters reported on October 9, 2026 that India has about 4% of the world’s freshwater for 18% of its population. It also cited an estimate that 60% to 80% of Indian data centres could face high water stress this decade.
Local authorities should ask:
- Will cooling compete with drinking water during dry periods?
- Can recycled or non-potable water be used instead?
- Who funds new pipelines and drought safeguards?
Thane Has Turned the Debate Into a Neighbourhood Fight
More than 1,000 residents near a planned 54-acre Amazon data-centre project in Thane have raised concerns over water, electricity and noise. Amazon says the facility will use water-free cooling and dedicated power rather than household supplies.
That disagreement is why disclosure belongs before construction, not after residents discover a project’s scale beside their homes.
Land Is the Cost AI Companies Rarely Advertise
AI infrastructure needs more than server halls. Large campuses can require substations, transmission corridors, roads and security buffers. That can change neighbourhoods, compete with farmland or affect wildlife and public access.
The argument is spreading. A US conservation group called for a moratorium on AI data centres on federal public land, citing water, electricity, wildlife and access concerns. Spain is also moving toward tighter oversight of data-centre electricity .
These disputes do not prove data centres are automatically bad neighbours. They show that “jobs and investment” cannot be the entire approval test. Land-use decisions need consultation, local benefits and enforceable resource plans.
The AI Boom Needs a Developer-Pays Rule
Governments want computing capacity because AI has economic and strategic value.
Developers should fund grid connections, utility upgrades and resource mitigation linked to their projects. Regulators should publish water and power commitments before approvals. Local authorities should seek community-benefit agreements where large campuses reshape surrounding areas.
Water-free cooling, recycled water, cleaner power and efficient chips can reduce pressure. Yet efficiency cannot replace transparent accounting while facilities keep getting larger.
The AI boom is often presented as weightless progress. It is not. It has transformers, pipes, acres and electricity meters. Governments can welcome innovation without treating its physical footprint as somebody else’s bill.
Why do AI data centres use so much electricity?
AI servers run powerful specialised chips continuously, while cooling, networking and backup equipment add further demand, particularly in facilities handling model training and inference daily.
How much water do data centres consume?
Water use varies by cooling technology and climate. US data centres directly consumed about 66 billion litres in 2023, according to Lawrence Berkeley National Laboratory
Can AI data centres increase electricity bills?
They can raise system costs when utilities need new generation, substations or transmission and those expenses are shared with ordinary customers instead of developers over time.
Why are communities opposing new data centres?
Residents commonly raise concerns about water availability, electricity demand, noise, land use, construction traffic and whether promised economic benefits justify local infrastructure pressure over time.
What should governments require before approving AI data centres?
Authorities should require transparent power and water estimates, infrastructure funding plans, environmental review, local consultation and clear responsibility for costs created by unusually large loads
The Hidden Cost of the AI Revolution
Can India Sustain Its AI Expansion?
Why Are Indian Data Centres Expanding?
Find out how Bhubaneswar is emerging as a destination for AI computing and cloud infrastructure
Could Nuclear Energy Power AI Growth?
Can Renewable Energy Support AI Infrastructure?
See how Abu Dhabi is connecting AI development with renewable power and energy infrastructure.
Are Global Tech Giants Expanding Datacentres?
