INDIA’S FIRST PRIME MINISTER Jawaharlal Nehru in a letter to Chief Ministers on August 15, 1957 wrote, “We have many large-scale river valley projects which are carefully worked out by our engineers. I wonder, however, how much thought is given before the project is launched, to having an ecological survey of the area and to find out what the effect would be on the drainage system or to the flora and fauna of that area. It would be desirable to have such an ecological survey of these areas before the project is launched and thus to avoid an imbalance of nature.” Those thoughts were expressed by him after river valley projects were built and it is instructive to note that he underlined the critical necessity of making environmental assessment before setting up such projects.
In the context of the growing ecological crises, and intensification of climate chaos and global warming those articulations of Nehru assume significance to assess the environmental impact of Artificial Intelligence Data Centres before these are set up. It is especially so because in large parts of the world widespread concerns have been expressed about the extremely adverse impacts of such Data Centres on environment and their possible role in exacerbating the crises of climate change and global warming. One particular proposed legislation in USA titled “Artificial Intelligence Data Center Moratorium Act’’ has called for imposition of moratorium on the construction of new data centers until legislation is enacted by the American Congress safeguarding the public from the dangers of artificial intelligence.
The Issue of Data Centre in Vizag
Against this backdrop it is important to see and evaluate Google’s planned AI hub in India, developed with partners including AdaniConneX and Airtel. It claims to promise computing infrastructure, international connectivity, and a larger role for India in the digital economy. Google announced approximately $15 billion in investment for 2026-2030 in Visakhapatnam, with construction already underway in April 2026, and promises new transmission infrastructure, clean electricity generation, and energy storage. Though it claims that it would boost Andhra Pradesh’s economy, the infrastructure behind this project requiring vast swathes of land and enormous quantity of electricity and water generated extreme worry and anxiety about its horrible consequences on ecology of the region. The question here is not whether India can attract investment, but, equally consequential, whether the present environmental regulations can assess the impacts of these projects and how they would affect the local ecology throughout their operating lives. This has already been highlighted by the Human Rights Forum (HRF), which has challenged the ecological suitability of the Adavivaram–Mudasarlova site, citing its proximity to a drinking-water reservoir and the Kambalakonda Wildlife Sanctuary, and has raised objections regarding both the location and the adequacy of the environmental appraisal. While there is nothing to indicate that the current regulations have not been followed, the question is whether those approvals adequately capture the authorised infrastructure and the choices being made for the people in and around the area.
The Adavivaram–Mudasarlova data centre’s environmental significance cannot be assessed solely from the floor plan or the environmental clearances. The environmental clearance and no-objection from the Forest Department were issued on 10th June 2026, describing a 512 MW data centre with approximately 106,027 square metres of built-up area across 64.75 hectares, and additionally identifies a water requirement of 224 kilolitres per day and 96 diesel generators providing approximately 301 MW of backup capacity. However, even with the approvals, broader questions of electricity supply, emergency emissions, and resource allocation, especially in an eco-sensitive region, are raised. In fact, questions about the rush of these clearances are also at the heart of this controversy and warrant closer inspection. Even though the recent Environmental Impact Assessment (EIA) notification allows for category B2 projects to be approved without a full EIA Report, including exemptions surrounding public consultation and other participation requirements. However, specifically for AI infrastructure, especially given the reports of intense dependence on local resources, a thorough assessment should also evaluate the environmental burden, including the intensity and duration of resource use.
Relevance of SC’s Vanashakti Judgment
The relevance of EIA clearances has recently been evaluated repeatedly in the Vanashakti judgments and reinforces a principle central to the AI infrastructure debate: where prior environmental clearance is required, it must pre-date the construction, and the Supreme Court upheld a narrowly tailored time-bound 2017 regularisation notification but prospectively quashed the 2021 Office Memorandum, which had created a permanent route for projects to seek clearance after commencing without it. An administrative instruction could not displace the mandatory requirements of the 2006 Environmental Impact Assessment Notification. The 2025 judgment does not impose an absolute prohibition on post-facto clearance, and existing clearances remain valid unless individually challenged, while simultaneously preserving the government’s power to issue narrowly tailored amnesty notifications in ‘supervening public interest’ and its own power to grant exceptional relief. The 2026 judgment also complements this reasoning, and in striking down exemptions for industrial sheds and educational buildings, the Court found that their intended use did not justify excluding substantial construction from environmental scrutiny, especially an assessment by an expert body such as the State Environment Impact Assessment Authority.
These precedents are quite informative regarding AI infrastructure, especially when post-construction assessment is inadequate. When a project exists on the ground, appraisal cannot recreate the opportunity to compare alternative sites or even enable public consultation about the environmental impacts that would have shaped the approvals. Applied to data centres, this supports the argument that promises of innovation, investment and digital growth should inform environmental assessment without replacing it. For the proposed Visakhapatnam project, the chronology therefore matters: which activities occurred, when, and which approvals were legally required before each activity began? Allegations of premature construction or land preparation must be tested against that record. A later clearance cannot, by itself, establish that earlier work was lawful. The EIA system was incorporated with a view to being preventive and not merely a regulatory requirement. Environmental scrutiny must retain the ability to influence a project’s location, design and resource demands before cleared vegetation, altered terrain, and installed infrastructure make those decisions difficult to reverse. Hence, each element of the environment needs closer inspection, given that there has been no official assessment yet, with clearances being rushed, and concerns raised by local communities.
Opposition to Data Centre
There has been staunch opposition and recorded concerns from Dalit farmers and communities in Tarluvada over land acquisition, compensation, and livelihoods, highlighting competing claims over land. Whereas adequate compensation can address part of a loss, but its adequacy cannot be judged only against its market price, it also needs to take into account cultivated land, which provides a recurring income, household security, and access to work for people who do not directly hold title themselves, along with the large-scale displacement of indigenous communities. In fact, clearing vegetation and altering slopes can change runoff and sediment movement in the region where a site is connected to a reservoir catchment. This brings us to the second and most pressing issue for AI data centres – water, given concerns about cooling and construction demands. Google has maintained that the Visakhapatnam project will use advanced air cooling, and the Government of Andhra Pradesh has said that rural and residential water supplies, as well as the nearby reservoir, will not supply the forthcoming data centres. However, what happens during an exceptionally dry month or season? Seasonal reliability matters because industrial demand is being added to an existing landscape of household, agricultural, and ecological needs, as India is, overall, a rain-fed agrarian economy (60%). An assessment would have been helpful in determining which uses receive priority when supplies fall short of expectations, so as not to surprise residents by discovering those priorities only after infrastructure becomes operational and service is reduced.
Energy Guzzling Data Centre
Another aspect of these AI data centres would be the scale of electricity demand, not merely during the construction but during its operations as well. The International Energy Agency’s updated 2026 analysis projects that global data-centre electricity consumption will rise from approximately 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030. It also identifies rapid growth in AI-focused facilities and challenges associated with supplying their changing loads. Google’s commitments to clean generation, transmission, and storage are relevant; however, they require alignment with state and national priorities and infrastructure, for which Google alone cannot be held responsible. Their environmental value depends on delivery: how additional capacity meets demand throughout the day, how backup generators and their emissions compare with installed capacity, the financial question of who bears responsibility for additional electricity supply, and the neighbourhood exposure.
The Visakhapatnam hub comprises several sites and project entities that may be large enough to be classified as category A or B1 under the EIA notification. Just because they are considered separate legal entities does not establish that their environmental demands are independent. This is the case for cumulative assessment. It requires authorities to ask whether separate approvals, taken together, remain compatible with environmental capacity. Criticism of the approval process must be legally precise and should be raised sooner rather than later. Proximity to a sanctuary does not, by itself, establish that a construction proposal necessarily requires Category A appraisal. The current consolidated notification contains exceptions concerning the application of its General Condition to item 8 projects. Proximity to protected areas and ecologically sensitive locations remains important and should be assessed in proportion to their risks through the current administrative route, but the legal conclusion requires examination of the applicable provisions and orders. An important tenet of the legal argument could be centred on public consultation or environmental participation, allowing residents to identify risks that mere technical submissions may overlook – such as seasonal shortages, drainage routes, participation in cultivation, ecological proximity, and dependence on shared resources. Meaningful consultation requires accessible information while alternatives remain possible, and a public meeting after land commitments and major investment decisions has little influence if the location and design are already considered settled. Ideally, authorities should publish applications, appraisal reasons, relevant maps, and supply commitments in a form that residents can understand, specifically before such projects are sanctioned, not as an afterthought.
Necessity for a Regulatory Approach
India, ideally, needs a regulatory approach that treats high-density computing as a continuing, resource-intensive infrastructure, especially in the face of climate change, and that ensures these obligations extend through expansion, equipment replacement, and eventual decommissioning. A specific appraisal framework for large data centres could combine electricity demand, land area, water requirements, ecological sensitivity, and the overall environmental footprint. Operators should disclose resource budgets before approval and report actual performance afterwards, and disclosure should distinguish freshwater withdrawal from consumption, installed electrical capacity from actual use, and renewable procurement from operational supply. Independent verification would make these distinctions useful.
Example of European Union Relevant for India
Other jurisdictions offer examples rather than ready-made solutions, and not as a completed regulatory model. The European Union has introduced data-centre energy-performance reporting and a database covering relevant energy and water indicators. The European Commission proposed a common rating scheme in September 2026 and is developing minimum-performance proposals. Learning from these schemes, India can adapt the principle of measurable accountability to its own conditions. Drought-sensitive supply rules, enforceable operating limits, community grievance mechanisms and periodic review should accompany investment commitments. Subsidies should depend on demonstrated performance, with consequences where promised safeguards are not delivered. The Google–Adani hub presents an opportunity to establish those expectations before further projects follow, especially in other eco-sensitive areas such as the Nicobar and Lakshadweep Islands, and reliable infrastructure benefits should emerge prior to its construction through the EIA, with clear rules, credible assessments, and communities that can see how decisions were reached. India’s AI ambitions will require choices about land, electricity and water. Those choices should be explicit, supported by evidence and open to public scrutiny. The environmental cost of computing cannot be left for neighbouring communities to calculate after the investment has been celebrated.
The author would like to thank Khusshi Mittal, a 4th-year student in the BA LLB Programme and Kavya Jiyyani, a 2nd-year student in the BA Legal Studies Programme at Jindal Global Law School, for their research assistance
