The Data Centre Deluge: Who Pays for the AI Revolution?

The Data Centre Deluge: Who Pays for the AI Revolution?

By Alan john Koshy

Across India and Aotearoa, the unchecked expansion of hyperscale data centres is meeting growing grassroots resistance.

From the streets of Thane and Visakhapatnam to the farmlands of Southland, communities are resisting an infrastructure boom that is imposing immense environmental and social costs on locals.

This is not just a local planning dispute. This is “technofeudalism”, driven by a handful of tech oligarchs with both immense wealth and a quasi-religious zeal for artificial intelligence.

The data centre is the physical manifestation of this new feudalism. Like some insatiable beast from ancient myth, it demands we feed it our precious water, land and energy, while offering almost nothing in return.

The Frontlines: From India to Aotearoa

In India, the government’s ambition to become an AI powerhouse is colliding with local reality.

In Thane, just north of Mumbai, locals are protesting a proposed 53-acre Amazon hyperscale data centre in a densely populated residential area, surrounded by housing complexes, schools, and hospitals.

Residents fear it will worsen existing water shortages, with some calculating the centre could require up to 12 million litres of water daily; enough to fill nearly 1,000 swimming pools every week.

Beyond water, residents fear strain on the electricity grid, noise pollution that could reach 90 decibels, and the immense heat emanating from the facility's cooling systems.

Nearly 800 kilometres away in Visakhapatnam, a similar battle is brewing against a proposed US$15 billion Google hyperscale AI complex, in partnership with Adani Group and Bharti Airtel. It is slated to be the tech giant's largest AI hub outside of the United States.

Here, the concerns are amplified by the project's proximity to the Kambalakonda Wildlife Sanctuary, less than a kilometre away, in a city already facing significant water stress.

India's data centre capacity has nearly tripled since 2020, from 520 MW to 1.5 GW, and is projected to hit 4.5-6.5 GW by 2030. Wood Mackenzie forecasts it to hit a whopping 12 GW. Google, Amazon and Microsoft have already pledged a combined US$68 billion for AI-driven projects across India.

In Aotearoa, the Stop Data Centres NZ movement is organising protests, calling for a national framework and a pause on new hyperscale developments. Their petition demands national oversight of effects on water supplies, the electricity grid and productive land.

The planned Datagrid facility in Makarewa, Southland, with its 240-megawatt capacity and heavy groundwater use, has become a key battleground.

The Thirsty Subcontinent

To understand the fury of residents in Thane and Visakhapatnam, one must first grasp the scale of India's water crisis. It is a catastrophe born from a warming planet and decades of unchecked urban expansion.

India is home to 18% of the world's population but has access to a mere 4% of its freshwater. The country is the world's largest groundwater extractor, accounting for over 25% of global withdrawals.

An alarming NITI Aayog report warns that by 2030, 40% of India's population—about 600 million people—may have no access to drinking water.

Chennai ran dry in 2019, becoming India's first "Day Zero" city. Bengaluru once had over 260 lakes; today, only 80 remain. Delhi's 33 million people often survive on thousands of water tankers. Hyderabad, the most groundwater-stressed region, faces a shortfall of 909 million litres a day. Yet Amazon is expanding its data centre there, while Microsoft plans to expand in Pune, where water shortages sparked protests last year.

Nearly 75% of India's data centres are located in regions facing high to extremely high water stress.

Despite this, the union and state governments are offering massive tax holidays and fast-tracked approvals for these water-guzzling facilities, while their citizens struggle to secure their daily water supply.

The Hidden Costs

The environmental and social costs that communities are protesting are staggering:

  • Energy and Emissions: The largest data centres use as much electricity as 2 million households combined. This demand is extending the lives of dirty coal and gas-fired power plants, undermining the transition to renewables.
  • Water Consumption: If servers overheat, they fail. A single hyperscale data centre can consume millions of gallons of water daily for cooling, often drawing from stressed local supplies. This water is effectively lost forever, evaporating or becoming too contaminated to return to the system.
  • Land Use and Displacement: These sprawling, industrial complexes lock vast tracts of land into permanent use, displacing housing, agriculture, and green spaces. Yet even the largest centres often employ fewer than 150 staff, while the AI they power threatens to automate entire industries.
  • E-Waste and Toxic Legacy: Servers have a short lifespan, fuelling a global e-waste crisis. Less than a quarter is recycled; much is dumped on developing countries, exposing communities to lead, mercury and cadmium. Cooling systems are laced with PFAS "forever chemicals" that accumulate in soil and water and are linked to cancer, immune suppression, birth defects and high cholesterol.
  • Noise: Local communities have compared the constant buzzing of data centres to "lawn mowers running 24 hours a day," seeping through closed windows and eroding quality of life.

These impacts explain why AI data centres are increasingly located in countries like India, Australia, Ireland, Chile and Brazil, regions with weaker environmental regulations or cheaper resources, while the profits flow back to Silicon Valley.

In the US, a Gallup poll found that Americans would rather live next to a nuclear reactor than an AI data centre.

This is environmental colonialism: communities pay the price, tech bros take the profits.

The fight is not against technology itself, but against a system that demands we sacrifice our communities on the altar of corporate profit.

The AI Bubble Could Pop Loudly

The fight against data centres is not just about water, land, and pollution. It's about a financial house of cards that could bring the global economy crashing down.

More than US$200 billion in data centre debt was raised in 2025 alone, with tech companies borrowing at levels that surpass the dot-com era, even after adjusting for inflation.

Much of this involves "off-balance-sheet financing, chip-backed collateral and complicated leasing structures", much like the murky financial engineering that triggered the 2008 crash.

What makes this bubble uniquely dangerous is the circular “closed-loop” financing system that props it up. The entire structure depends on constant capital inflows. If credit conditions tighten or a major player stumbles, the system could topple like dominoes.

When the dot-com bubble burst between 2000 and 2002, the impact was largely limited to stock market losses. But when a credit bubble bursts, assets are sold off at fire-sale prices, and a wave of bankruptcies wrecks the banking industry and the global financial system.

The awkward reality? The companies building AI are not profitable. OpenAI may burn more than US$140 billion before turning profitable, nearly five times more than Amazon, Tesla, and Uber's cumulative early losses combined. An MIT study found that 95% of companies see zero return on their generative-AI investments.

Goldman Sachs warns that US$19 trillion in market value is detached from economic reality. By comparison, the dot-com bubble looks muted.

None of this adds up. The debt is unsustainable, the returns are illusory, and the crash is inevitable. So why is Big Tech ploughing ahead, indifferent to the economic logic? The answer lies not in economics, but in ideology.

The Technofeudal Vision

Under technofeudalism, a handful of tech oligarchs control the digital infrastructure on which modern life depends. They extract "cloud rents" from businesses and governments that become dependent on their platforms, creating a new form of digital serfdom.

The data centre is the stronghold from which these lords extract their tribute. Like the castles of old, they are built on land seized from communities and defended by armies of lawyers.

But unlike the old lords, they do not live among the people they rule. They reside in distant mansions and superyachts, accumulating wealth that would make medieval kings blush.

As investigative journalist Whitney Webb has documented, many of these figures hold a "quasi-religious fervour" for artificial general intelligence. They are openly transhumanist, believing humanity's destiny is to merge with AI.

This ideological zeal propels the data centre boom even when the economics make no sense. Computing power is their god, data centres are their cathedrals, and communities are the sacrifice.

Fighting Back: Strategies for Resistance

Across the world, communities are successfully fighting back against the data centre deluge. Their strategies offer a blueprint for resistance:

  • Start Early and Get Informed: Research local zoning laws, file Freedom of Information requests to uncover planning applications, and monitor changes in utilities to spot projects before they’re announced.
  • Build Broad Coalitions: The fight unites conservatives and progressives, church groups and environmentalists, local businesses and sports clubs.
  • Advocate for Strong Local Regulations: Push for ordinances that require community notice and input upfront, rather than allowing data centres “by-right” in any zone. “Special-use permits” force a front-end conversation about air, water, and noise.
  • Push for Moratoriums and Frameworks: A moratorium is not a radical demand; it’s a sensible pause to “reboot” and create time for proper regulation. Campaigns in Aotearoa and the U.S. are calling for frameworks to ensure transparency, sustainability, and grid stability.
  • Divestment and Boycott: By divesting from big tech products and services as much as possible, we can undermine the demand that justifies the data centre deluge.

We don't need more efficient data centres. We don't need better reporting standards. We don't need to make extraction more palatable. We need to stop building them altogether.

The protests in India, Aotearoa, and across the globe show the resistance has begun.

 

Alan John Koshy is a freelance writer and supporter of the Aotearoa Alliance of Progressive Indians. He is a former Reuters business journalist and holds a Master's in Economics from the University of Canterbury.