Indian Protests Did Not Stop the Data Centres: They Repriced Them

There are three numbers describing how much water one building in Thane will drink, and they do not agree with one another by a factor of more than a thousand.

The building is a hyperscale data centre that Amazon Data Services India is developing on roughly 54 acres in Balkum-Majiwada, land bought from the Kalpataru Group in October 2022. The first number comes from Amazon's own public fact sheet, which puts the approved municipal water supply permit at 0.01 million litres a day — ten thousand litres, strictly for restrooms and drinking water. The second comes from filings made to the State Level Expert Appraisal Committee in August 2025, which described a requirement of 125 kilolitres a day; Amazon's later accounting reconciles this as ten kilolitres for human needs plus 115 kilolitres of treated sewage water for landscaping and irrigation. The third number comes from the Thane Municipal Corporation, which residents say has allocated 12 million litres a day of recycled sewage water to the facility.

Ten thousand litres. One hundred and twenty-five thousand. Twelve million.

The residents' group WakeUp Thanekars put those three figures side by side and asked the obvious question, which is not “how much water will this use” but “why does nobody appear to know”. That is the question that has been travelling across India for the past several months, and it is the question that, according to a Communications Today report published on 24 August 2026, has coincided with foreign funding for Indian data centres falling from a range of 600 to 900 million dollars a month in the first half of the year to roughly 100 million, with no foreign investment recorded at all in August.

Zero is a striking number. It is also, on its own, a nearly meaningless one. What follows is an attempt to work out what it actually measures.

A Building That Was Never Supposed to Be Controversial

Balkum was not chosen for a fight. It was chosen because Thane is forty minutes from the Mumbai metropolitan region that holds, according to CBRE, some 53 per cent of India's operational data centre supply, and because Maharashtra had made itself extremely easy to build in. The state's IT and ITeS policy offers up to 60 per cent exemption on electricity duty for fifteen years, full stamp duty exemption on land purchased for a data centre, additional floor space index, and single-window clearance. The Maharashtra industries department has earmarked land parcels inside MIDC industrial areas as dedicated data centre parks, and the state has classified data centres as an essential service — which means that when water and power are short, the facility is served first.

That classification is the detail residents keep returning to, because Thane is already short. At the mass gathering on 18 July 2026, where more than a hundred residents formed a human chain, Dayanand Nene noted that the city runs a daily shortfall of around 30 to 31 million litres. Meghna Shetty, an architect involved in the campaign, said that when representatives met the group they seemed to have understood the concerns but had no answers. Jatin Kothare, a finance professional, pointed to the 4,500 air-conditioning units planned for the site. Anuradha Rode, who lives in an adjacent residential complex, described the prospect of heat rejection from a twenty-two-storey structure reaching villagers' houses, and told reporters that the facility belonged in a designated technology park, not forty metres from bedroom windows. Manoj Pradhan, another opponent, framed it as a category error: not a corporate office block but industrial-scale digital infrastructure dropped into a densely packed residential ecosystem.

Amazon's rebuttals are specific and, in their own terms, credible. The company says it uses no water for cooling at any of its Indian data centres. It says the facility will draw power through its own dedicated extra-high-voltage substation connected to the transmission grid, not the local distribution network that serves homes and shops. It says the environmental impact assessment is publicly available on the government's PARIVESH portal, that tree felling was conducted under a valid Thane Municipal Corporation permit — 239 trees felled, 77 replanted, 2,209 compensatory saplings — and that generators run mostly for brief mid-day tests. It has invested 1.2 million dollars through its Right Now Climate Fund in Thane Creek restoration and funds STEM education for 900 students across nine schools.

Both accounts can be substantially accurate at once, and that is precisely the problem. A facility that consumes almost no water for cooling can still hold a 12 MLD municipal allocation on paper. A facility drawing from the transmission grid still draws 422 megawatts from a system that serves everybody. A facility with 193 diesel generator sets on standby still has 193 diesel generator sets. Nene's argument, as reported, was not that the numbers were sinister but that a project of this scale ought to be developed with adequate transparency, proper land-use compliance, infrastructure planning, and protection of the interests of the people living around it — and that the documentary record did not demonstrate any of those things.

The residents wrote to Chief Minister Devendra Fadnavis. They met the municipal commissioner, Saurabh Rao. They sent legal interrogatories. They said they would go to the National Green Tribunal and the Bombay High Court. Flat prices in neighbouring complexes, by their account, fell by 20 to 50 lakh rupees.

An Index That Measures the Right Thing

Two weeks before the Balkum gathering, a review paper appeared on arXiv that supplies the analytical vocabulary this dispute has been missing. “AI Data Centers and the Water Use Feedback Loop”, by Basit A. Akinade, Amobichukwu C. Amanambu, Jonathan M. Frame and Shaolei Ren, submitted on 19 June 2026, formalises what the authors call the Water and AI Feedback Loop: data centres consume water for cooling, water scarcity constrains where they can be sited, and AI tools are in turn sold as a means of improving water system efficiency. These three dynamics are usually studied in isolation. The paper's contribution is to insist they are one circuit.

Its practical instrument is the Water Consumption Impact index, which quantifies a facility's demand not against national or basin-level totals but against the capacity of the utility that actually has to serve it. Applied across ten United States sites, the index produces a range that spans three orders of magnitude: from 0.2 per cent of host capacity to 134 per cent.

That upper figure deserves a moment. A Water Consumption Impact above 100 per cent means the facility's demand exceeds the entire capacity of the public water system hosting it. In national terms it may still be a rounding error — data centres remain a small fraction of any large country's total withdrawals. In community terms it is the whole system and then some.

This is the paper's central claim, and it reframes the Indian argument completely: the AI and water problem is not a national volumetric problem, it is a community-scale utility burden problem. Aggregate reassurance is therefore not merely unhelpful but structurally misleading. When Andhra Pradesh officials say no rural or residential water will be diverted, or when Amazon says its Indian facilities consume no cooling water, both statements can be true at the level they are pitched while telling a resident of Balkum or Adavivaram nothing whatsoever about the pipe outside their house. The index makes the mismatch legible. It also explains why every one of these disputes turns into an argument about numbers that nobody can reconcile: the parties are measuring against different denominators, and only one of those denominators is the one people live inside.

A companion review submitted the same day, “AI Data Centers and Power System Sustainability” by Yuhao Huang, Novarun Deb and Hamidreza Zareipour, runs the same logic through the electricity grid. Its finding is that rapid, spatially concentrated data centre load growth is outpacing clean energy deployment in several major regions — that the problem is not the total terawatt-hours but where they land and how fast. Concentration, again, is the variable that matters. Both papers arrive at the same structural conclusion from opposite utilities.

The Reservoir That Was Already Short

If Balkum is the case where the numbers do not reconcile, Visakhapatnam is the case where they do, and the result is worse.

Vizag receives roughly 410 million litres of water a day from its reservoirs and rivers against a stated requirement of 480 million. Rationing is routine for its 2.5 million residents. Into this, in October 2025, Andhra Pradesh signed a memorandum with Google's subsidiary Raiden Infotech for a gigawatt-scale AI hub, part of a 15 billion dollar commitment, with Adani Infra designated as the notified partner and AdaniConneX and Airtel's Nxtra as build partners. The state allocated 601.4 acres across three sites: 266.6 acres at Tarluvada near Anandapuram, 160 acres at Adavivaram and Mudasarlova, and 175 acres at Rambilli in Anakapalli district. The projected employment figure attached to the announcement was 188,000 jobs.

The objections are not primarily about volumes. They are about process. The 160-acre parcel at Adavivaram and Mudasarlova is endowment land belonging to the Simhachalam Devasthanam, and a public interest litigation now challenges its allocation. Land Conflict Watch's documentation records that around 204 acres of the Tarluvada holding consist of unrecorded “Shivaijamindar” lands whose ownership is undocumented, that survey numbers do not match actual possession, and that transfers of assigned land may run against the Andhra Pradesh Assigned Lands (Prohibition of Transfers) Act, 1977. Fifty-one households, 244 people, were recorded as affected as of February 2026. Compensation for D-Patta land was revised from 17 to 20 lakh rupees an acre after farmer representations, and for unrecorded land from 8.5 to 10 lakh. Many of the affected landholders are Dalit; one described having cleared jungle by hand over fifteen years to make the land workable.

The environmental record is where the compliance question becomes sharpest. The project was classified Category B2, a designation that exempts it from a full environmental impact assessment and from the public consultation that would otherwise accompany one. Around 90 per cent of the Tarluvada site reportedly falls within the Pedda Chukka Konda Reserve Forest. The Kambalakonda Wildlife Sanctuary sits 860 metres away. The Human Rights Forum issued an alert on procedural violations on 18 October 2025, a week after the initial allocation order, and has since pursued matters before the National Green Tribunal's southern bench. The activist group Jal Biradari brought the water case; the Andhra Pradesh High Court asked the state government to respond, and the litigation was listed for hearing on 24 August 2026 — the same day the investment figures were published.

Andhra Pradesh's position is that no rural or residential water will be diverted, and that risk assessment has not been skipped. Google says it will deploy advanced air cooling specifically to protect local water resources. Neither statement addresses the objection actually being made, which is that the paperwork establishing who owns the land, what the cumulative environmental burden is, and which reservoir will supply what, was not completed before the land was handed over. In July 2026, residents and activists marched carrying placards reading “We cannot drink DATA” and “Power for people, not for megabytes”.

Why the Paperwork Does Not Ask

The reason these projects can advance without answering the questions is not evasion. It is that Indian environmental law, as currently written, largely does not pose them.

On 6 August 2026, in a written reply to Parliament, the Union Ministry of Environment, Forest and Climate Change stated plainly that AI data centres do not, per se, require environmental clearance under the Environment Impact Assessment Notification, 2006. Clearance is triggered only if the facility forms part of a building or construction project crossing threshold areas — 20,000 square metres of built-up area, or 50 hectares of township development, or 150,000 square metres. The trigger is floor area. It is not megawatts, not litres per day, not tonnes of heat rejected, not decibels at the boundary. A facility can therefore be assessed on the basis of how much ground it covers rather than how much of a city's utility capacity it absorbs, which is exactly the denominator error the Water Consumption Impact index was designed to expose.

The rest of the regime is fragmented by design. State pollution control boards enforce standards under the Water Act of 1974 and the Air Act of 1981. The Ministry of Jal Shakti oversees groundwater extraction. The Bureau of Energy Efficiency has issued District Cooling Guidelines. The Ministry of Electronics and Information Technology has convened a working group on energy efficiency and water management in digital infrastructure. Nowhere in that architecture does a single authority hold a consolidated view of what a given campus will draw from a given system. The consolidated view could not be assembled in any case. A Down To Earth investigation of 21 August 2026 found that none of the major state data centre policies requires operators to disclose water consumption through a centralised reporting system, leaving regulators unable to identify emerging risks. It put national data centre consumption at around 150 billion litres a year, projected to more than double by the end of the decade, and cited Planet Tracker data placing India second among Asian countries for data centres in areas of extremely high water stress.

Bengaluru shows what happens when that architecture meets a decade of build-out. Karnataka has 32 private data centres, 31 of them in Bengaluru, clustered in Whitefield, Electronic City and the corridor towards Devanahalli. Nine major operators have committed 24,253 crore rupees for roughly 2,694 jobs. The state government's informal estimate of their combined water use is about 4,000 kilolitres a day; independent calculation, working from installed capacity and the standard planning figure, puts it nearer 20 million litres. The gap exists because, as a water consultant who examined the system observed, the Bangalore Water Supply and Sewerage Board does not monitor where water goes for commercial purposes, and no agency verifies after approval whether a declared water source matches actual practice. Meanwhile the Central Ground Water Board finds Bengaluru extracting close to ten times its annual recharge, and Devanahalli exceeding permissible extraction by 169 per cent. In 2025, domestic users in the city faced 5,000-rupee fines for non-essential water use. Data centres faced no equivalent restriction, because they are classified as IT and ITeS rather than as large infrastructure.

The national backdrop is not reassuring either. The Dynamic Groundwater Resource Assessment released by the Ministry of Jal Shakti in December 2025 found 730 of India's 6,762 assessment units — 10.8 per cent — categorised as over-exploited, with the national stage of extraction at 60.63 per cent and the over-exploited units concentrated in the north-west, west and south. NITI Aayog's Composite Water Management Index work has long put the number of Indians facing high to extreme water stress at around 600 million.

Against that, in the Union Budget presented in February 2026, India offered foreign cloud providers a twenty-year exemption running from FY27 to FY47 on global income routed through Indian data centres, and the Taxation and Other Laws (Amendment) Bill, 2026 moved the incentive from entity-specific approval to a condition-based framework, extending it to leased facilities. The country has built a fiscal regime of unusual generosity on top of an environmental regime that does not ask the relevant questions.

The Bill That Arrives on Someone Else's Meter

The electricity story is subtler than the water story, and in the long run probably more consequential.

Amazon's defence in Thane — that the facility connects to the high-voltage transmission grid through its own dedicated substation and will not draw from the local distribution network — is technically sound and rhetorically incomplete. Indian retail tariffs are built on cross-subsidy. Commercial and industrial consumers pay somewhere between 30 and 80 per cent above the cost of supply; agricultural consumers pay less than 20 per cent of it; residential tariffs sit below cost. Industrial users typically carry a cross-subsidy premium of two to four rupees a unit on top of average supply cost, and that premium is what keeps power affordable for roughly 250 million residential consumers and 150 million farming households. Commercial and industrial demand is 40 to 43 per cent of the total.

A data centre is, on paper, the best customer a distribution company could hope for: a flat load, high consumption, reliable payment, tariffs of ten to twelve rupees a unit. But data centres are also unusually well placed to leave. Open access power purchased through long-term renewable contracts runs at roughly 3.50 to 5.00 rupees a unit against commercial discom tariffs of eight to eleven in Maharashtra and Karnataka, and the economics of a 400-megawatt load make the switch close to inevitable. Andhra Pradesh has gone further, approving private power distribution licences for data centre operators — an instrument previously reserved for utilities such as Tata Power and Adani Electricity Mumbai — allowing a hyperscale campus to become, in effect, its own discom, procuring directly and even sub-distributing unused contracted power.

Each of those steps is defensible on its own. Together they describe a mechanism by which the largest new industrial loads in the country exit the pool that subsidises everyone else's electricity, while the fixed costs of the network they were connected to remain to be recovered from a smaller base. The residents of Balkum worrying about power cuts are worrying about the wrong mechanism. The bill does not arrive as an outage. It arrives as a tariff order.

What the Investors Started Writing Down

Which brings us back to the number that prompted this article.

According to the Communications Today report, the deceleration was not only in value but in frequency: from two or three deals a month until June to one a month afterwards. It found that project documents showed discrepancies between the documented sourcing and level of water usage and the figures quoted publicly — the Balkum pattern, generalised. And it found that investors had begun raising community questions directly and increasingly in writing, with diligence lists now carrying the specific campus water source, the cooling architecture and the power arrangement as named line items.

That last detail is the one worth dwelling on, because it is the only part of this episode that is structural rather than episodic.

A protest is a temporary fact. A written diligence question is a permanent one. Once “what is the specific water source for this campus, and who else draws on it” enters a standard investment memorandum, it stays there for every subsequent deal in every subsequent state, and it stays there whether or not anybody is still marching in Thane. It converts a community grievance into a financing condition — which is to say, into a risk that has a price and therefore a lawyer. The same conversion is visible in independent advisory work on Indian data centre investment, which now treats grid capacity verification at 220 and 400 kV substations, thermal cooling architecture and state-specific permitting workflows as the difference between a performing asset and a stranded one, rather than as compliance boilerplate.

State policy has moved in the same direction and faster than the capital has. Gujarat notified the Viksit Gujarat Data Centre Policy 2026-29, targeting up to 7.5 gigawatts of capacity and an anticipated six lakh crore rupees of investment, and it did so with two hard physical conditions attached: eligible projects must source at least 51 per cent of operational electricity from renewables, and must draw water from desalination rather than from freshwater meant for agriculture or drinking. Dholera Industrial City Development issued a tender for consultants on a 200 MLD seawater desalination plant on 9 July 2026. Karnataka's information technology minister, Priyank Kharge, told the state assembly that data centres were heavy water and energy guzzlers, cited a planning figure of 25 million litres of water per megawatt per year, and confirmed the state's two-to-three-year-old data centre policy was under review with a greener framework and a push towards Mangaluru, Mysuru, Hubballi-Dharwad, Kalaburagi, Shivamogga and Tumakuru.

These are not gestures. A desalination mandate is a capital expenditure line and a physical constraint. A 51 per cent renewable floor is enforceable. Neither would have been written in the absence of the protests.

What the Precedents Actually Show

India is not the first place to run this experiment, and the international record is more instructive than either side of the argument tends to allow.

In Uruguay, Google's proposed facility in Canelones, the department supplying Montevideo, was designed to consume around 7.6 million litres of water a day — the domestic consumption of some 55,000 people — during the country's worst drought in 74 years, at a moment when authorities were mixing saltwater into the public supply. The protest slogan was “it is not drought, it is pillage”. Uruguay's environmental authorities subsequently approved amendments moving the facility to air cooling. The project was not cancelled. It was reformulated, and academics promptly pointed out that the water saving was purchased with additional electricity, which is a transfer between utilities rather than a reduction in burden.

In Chile, Google's 200-million-dollar Cerrillos facility in Santiago received initial authorisation in 2020, then drew sustained opposition over the capital's strained aquifer after litigation revealed the cooling towers would require 7.6 million litres of potable water a day. An environmental court partially reversed the permit, requiring the application to be revised to account for climate change. In September 2024 Google renounced the 2020 permit altogether in order to redesign around water-saving technology.

The Netherlands went furthest earliest. Amsterdam and Haarlemmermeer imposed a moratorium on new data centre construction in July 2019 over grid and environmental strain. The constraint has since migrated from planning to physics: in early 2026 the grid operator TenneT wrote to companies in the Schiphol region pausing connection requests because of congestion that may persist to around 2035, and in April 2026 a Dutch judge ruled that the Haarlemmermeer grid was simply full, delaying a data centre connection.

Ireland is the cleanest case of all, because the state publishes both sides of the ledger. The Central Statistics Office found data centres consuming 22 per cent of all metered electricity in 2024 and 23 per cent in 2025. EirGrid's de facto moratorium on new Dublin-area connections ran for years. In December 2025 Ireland ended it — but on terms: any data centre seeking a connection must install on-site generation or battery storage capable of meeting its full demand, and applications outside greater Dublin are considered case by case.

Read together, these four cases describe the same trajectory, and it is not the one either camp claims. Opposition rarely cancels a facility. What it reliably does is change the terms on which the facility is built, and then embed those terms in a rule that applies to the next applicant who never protested at all. Air cooling in Uruguay, permit revision in Chile, self-generation in Ireland, grid congestion pricing in the Netherlands: in every case the concession outlived the campaign.

Relocation Is Not Reduction

The uncomfortable half of that finding is that terms which travel can also travel somewhere quieter.

The economics literature on burden displacement is old and its verdict is genuinely mixed. The pollution haven hypothesis — that firms in resource-intensive or polluting activities migrate towards jurisdictions with weaker regulation — has been tested for four decades with results that range from confirmation in specific sectors to null findings to the occasional “race to the top”, where inward investment raises host standards rather than lowering them. Early surveys concluded that compliance costs were generally too small relative to labour, logistics and market access to drive location decisions.

For data centres the calculus is unusual, because the inputs that generate the objection — water, power, land, grid headroom — are simultaneously the inputs that determine whether the facility can operate at all. That cuts both ways. It means a developer cannot simply relocate to a jurisdiction with lax rules if that jurisdiction also lacks water and transmission. But it also means that when a state does hold water and transmission, the incentive to site there rather than argue in Thane is very strong, and the community that happens to sit on top of those resources may have considerably less capacity to object than a Thane housing society with architects, finance professionals and a functioning residents' association.

Karnataka's revised policy pushing development from Bengaluru to Mangaluru, Mysuru, Kalaburagi and Shivamogga is, on the most generous reading, sensible load distribution. On a less generous one it is the relocation of a contested facility from a city with organised civic groups and national media attention to towns with neither. Gujarat's Dholera solution is stronger, because desalination is a genuine substitution of one water source for another rather than a transfer of the same burden to a different postcode — but it also concentrates enormous industrial development in a greenfield city where there is, as yet, almost no resident population to consult.

This is where the Water Consumption Impact index earns its keep as a policy instrument rather than merely an analytical one. If burden is community-scale, then the correct test for whether a relocation represents progress is not whether the new site is bigger or emptier but what fraction of the receiving utility's capacity the campus will absorb. A 200-megawatt facility on a coastal desalination supply and a 200-megawatt facility on a small municipal system are the same asset and completely different objects. Nothing in India's current approvals regime distinguishes between them, because the regime measures square metres.

What a Pause Is Worth

So: was August a victory?

Taken as a monthly data point, no. Monthly foreign investment flows into a lumpy, deal-driven sector are close to noise; a single large transaction slipping from one month to the next produces exactly this pattern. The structural evidence still runs the other way. CBRE recorded India crossing 1,700 megawatts of capacity in 2025 on 440 megawatts of new supply and projected roughly 30 per cent growth in 2026 on a further 500 megawatts. Cumulative commitments were reported at around 126 billion dollars by the end of 2025. Amazon, Microsoft and Google have collectively pledged tens of billions more. The tax holiday runs to 2047. On 3 September 2026, after the zero month, Commerce and Industry Minister Piyush Goyal said India had the potential to become the world's leading destination for data centres, and put the investment opportunity at more than 200 billion dollars against India's stated aspiration of a 30 trillion dollar economy by 2047. Nothing in a zero-dollar August alters any of that.

The sector's own data cannot agree on the baseline. Tracxn figures reported in July 2026 put foreign investor funding for India's data centres at 738 million dollars for the whole of the first half of 2026, with Nxtra alone raising close to 710 million from Carlyle Group, Alpha Wave Global and Anchorage Capital Partners. That is a run-rate of about 123 million a month, not 600 to 900. The two series count different things, disclosed equity rounds against broader capital commitment, and neither source reconciles them. A sector whose own trade data cannot agree within a factor of five is one where a single zero month establishes very little.

Taken as a signal, the answer is more interesting. What appears to have stopped is not investment but a particular kind of investment: the kind that could be underwritten without asking where the water comes from. The Communications Today findings describe capital that has not left so much as slowed down to read the file — deal frequency halved, questions submitted in writing, water source and cooling architecture and power arrangement promoted from footnote to line item. If that is what happened, then the protests did not repel capital. They raised its information requirements, which is a different and more durable thing, because information requirements do not get bored and go home.

And yet it would be dishonest to call that a victory for Balkum or for Adavivaram, because the specific facilities those specific residents objected to are still coming. The Thane campus has its permissions. The Vizag project has its land, its notified partner, its foundation stone laid in April 2026 and a target commissioning of July 2028. What the residents secured was not the removal of a building. It was the insertion of a question into the process by which every subsequent building will be financed — a question they will not be present to ask, on behalf of people they will never meet, in places whose names are not yet in the file.

There is a version of this that ends well, and it is not utopian. It requires the Environment Impact Assessment Notification to acquire a trigger keyed to load and utility burden rather than floor area, so that a facility drawing 422 megawatts and a municipal water allocation is assessed as what it is. It requires that allocations be published — the single reform that would have prevented the entire Balkum dispute, since the argument there was never about whether 12 MLD was too much but about why three official documents disagreed. It requires post-approval verification of the kind Bengaluru conspicuously lacks, where somebody checks whether the declared source is the actual source. It requires that downstream and neighbouring users have standing before land is handed over rather than afterwards in a High Court. None of these is technically hard. All of them are cheaper than litigation, and considerably cheaper than a month with no money in it.

The residents of Balkum did not set out to move the price of Indian digital infrastructure. They set out to find out how much water the building at the end of their road was going to use, and discovered that the answer existed in three incompatible versions across three official documents, none of which had been reconciled before the land was cleared and the trees came down. The Water Consumption Impact index found burdens spanning three orders of magnitude across ten American sites. Thane found a spread of the same magnitude within a single project's own paperwork.

Foreign investment can resume next month, and probably will. The three numbers still do not agree.

References

  1. AI Data Centers and the Water Use Feedback Loop – Basit A. Akinade, Amobichukwu C. Amanambu, Jonathan M. Frame and Shaolei Ren, arXiv preprint 2606.21760, 19 June 2026
  2. AI Data Centers and Power System Sustainability: Understanding the Sustainability Implications of AI-Driven Data Centers on Power Systems – Yuhao Huang, Novarun Deb and Hamidreza Zareipour, arXiv preprint 2606.21064, 19 June 2026, IEEE Energy Sustainability Magazine
  3. Communications Today – reporting on foreign funding, deal frequency and investor diligence in Indian data centres, 24 August 2026
  4. Amazon Defends Balkum Hyperscale Data Centre As Thanekars Demand Its Relocation – Free Press Journal, 18 July 2026
  5. 'Heat, Electricity, Noise...' Thanekars Continue to Fight Against Amazon Data Centre – Local Samosa, July 2026
  6. Amazon's data centre in Thane: Facts – Amazon India
  7. AI data centres do not need environment clearance, Centre tells Parliament – ThePrint, August 2026
  8. Environment Impact Assessment Notification, 2006 – Ministry of Environment, Forest and Climate Change, Government of India
  9. Concerns around acquisition process of Dalit farmers' land in the Google-Adani data centre project in Visakhapatnam – Land Conflict Watch
  10. Google's $15 billion India data centre project battles water, wildlife concerns – Reuters, 6 August 2026
  11. In Bengaluru's water-stressed areas sit most of its data centres – Newslaundry, 16 April 2026
  12. Karnataka's data centres 'guzzling water and energy', state reviewing its policy: Priyank Kharge in Assembly – Deccan Herald, 2026
  13. Viksit Gujarat Data Center Policy 2026-29 – Chief Minister's Office, Government of Gujarat, 2026
  14. Findings of Dynamic Ground Water Resources Assessment – Ministry of Jal Shakti, Press Information Bureau, Government of India, December 2025
  15. Composite Water Management Index – NITI Aayog, Government of India
  16. Incentives and Laws for Data Centers in Maharashtra – S&R Associates
  17. Andhra Pradesh Allows Private Discom Licence To Data Centre: Here's Why It Matters – The Core, 2026
  18. New Tax Laws Amendment Bill Clears Path For Simpler Data Centre Incentive Regime – Inc42, 2026
  19. India offers zero taxes through 2047 to lure global AI workloads – TechCrunch, 1 February 2026
  20. India data centre capacity to jump 30% in 2026; 500 MW supply boost expected: CBRE report – Business Today, 1 April 2026
  21. Google plans data centre in Uruguay, using millions of litres of water amid drought – Mongabay, 2023
  22. Chile partially reverses Google data center permit over water use concerns – Data Center Dynamics
  23. Judge rules electricity grid in Haarlemmermeer is full, data centre connection delayed – NL Times, 29 April 2026
  24. Ireland Ends Moratorium on New Power Links to Data Centers – Energy Connects, December 2025
  25. Data Centres Metered Electricity Consumption 2025, Key Findings – Central Statistics Office, Ireland
  26. FDI and environmental regulation: pollution haven or a race to the top? – Journal of Regulatory Economics, Springer
  27. India data centres attract $1.56 billion foreign capital – Tracxn data on H1 2026 foreign investor funding in Indian data centres, July 2026
  28. India Eyes $200 Billion Data Centre Investment – Channel iam, 3 September 2026
  29. India's AI Data Centre Boom Is Draining Water: Weak Regulation Threatens Long-Term Water Security – Down To Earth, 21 August 2026

Tim Green

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Listen to the free weekly SmarterArticles Podcast

Discuss...