States must assess power, water first

State governments should assess electricity supply, water availability and environmental impact before approving data centre investments, according to a new framework developed by NITI Aayog’s Frontier Tech Hub.

The Data Centre Evaluation Framework to Help States Plan for Sustainable Growth calls for a more rigorous assessment of proposed data centre projects. Rather than evaluating facilities primarily on their announced investment or computing capacity, the document recommends examining whether the necessary infrastructure will be available when a facility begins operating and whether the economic benefits justify the public support being offered.

The framework outlines five areas for evaluation: power and energy, water and cooling, siting and environmental sustainability, economic and strategic value, and policy implementation. It is intended to guide state-level decision-making rather than create an additional regulatory layer, according to NITI Aayog. Its recommendations would need to be adopted through relevant policies and approval processes before becoming binding requirements.

At the centre of the proposal is a shift in how states assess large digital infrastructure investments. The framework argues that the focus should move beyond competing for investment announcements towards demonstrating reliable infrastructure, sustainable resource use and measurable economic outcomes.

Power availability

One of the most significant recommendations concerns electricity supply, which is becoming a critical consideration as data centres expand their computing capacity to support artificial intelligence and other high-performance workloads.

The framework recommends that state governments evaluate the total electricity requirement of a proposed data centre campus, rather than considering individual buildings or phases in isolation. At the project planning stage, developers, electricity distribution companies and state transmission utilities should jointly prepare a Power Readiness and Ramp-up Plan, according to the document.

This plan would establish the total proposed load, sanctioned power capacity, expected utilisation, commissioning schedule and any transmission infrastructure required to support the development. States should also obtain load-flow studies covering existing and proposed electricity demand, transformer capacity, transmission constraints and contingency arrangements.

The framework recommends that approval for additional power consumption should be linked to the completion of the necessary infrastructure, rather than allowing developers to increase their load before the electricity network is ready.

It distinguishes between conventional colocation facilities and AI-focused data centres using graphics processing units, which are specialised chips designed for parallel computing tasks. These categories can have different power densities, cooling requirements and infrastructure needs.

For facilities exceeding 20 MW, the framework proposes automated demand-response readiness and integration with State Load Despatch Centre telemetry and control systems. Demand response allows large electricity consumers to adjust their consumption when the power system is under pressure. For data centres, this could involve managing suitable computing workloads or using available energy storage, without compromising critical operations.

The document also recommends assessing battery energy storage systems, microgrids and alternative backup power technologies to reduce dependence on diesel generators. For hyperscale and large computing facilities, it proposes an assessment of whether waste heat generated by computing equipment could be reused where technically and commercially feasible.

Water consumption

Beyond electricity, the framework identifies water availability as another important consideration in determining where data centres should be established. Large computing facilities generate substantial heat and require cooling infrastructure to maintain operating temperatures. Depending on the technology deployed, cooling systems may consume water directly or depend primarily on electricity.

The document recommends that states assess water requirements based on the proposed cooling technology, local climate and expected computing load. The assessment should consider peak summer conditions, including temperatures of approximately 40°C where applicable, rather than relying only on average annual consumption.

Before approving a project, the framework recommends coordination among water resources departments, urban local bodies, drinking water authorities, irrigation departments and industry representatives. These agencies should establish the water allocation available to the proposed facility, identify existing users whose supplies must be protected and determine how shortages would be managed.

The report calls for a long-term assessment of water availability over 20 to 30 years, taking into account seasonal variations, groundwater resources, surface water, treated wastewater and competing demand from agriculture. For locations already experiencing high water stress, states should require independent verification of water sources, certified meters, periodic reporting, rainwater harvesting and appropriate storage arrangements.

Where water scarcity makes additional freshwater extraction unsustainable, the framework recommends using alternative sources such as desalinated water rather than placing further pressure on existing resources.

The document makes an important distinction between proposed and assured supplies. Recycled or treated water should be counted towards a project’s available water only when the relevant supply infrastructure is contracted, connected and ready for use. This would prevent developers from relying on future water recycling arrangements that may not be operational when the data centre begins functioning.

Siting and economic value

The framework proposes a more systematic approach to selecting locations for new data centres. Instead of leaving developers to secure land and subsequently arrange electricity, water and telecommunications connectivity, states should identify locations where the essential infrastructure is already available or can be developed within a defined period.

Such data centre-ready zones would require advance assessment of land-use permissions, grid capacity, water availability, transmission infrastructure and access to multiple fibre network providers. The framework recommends that states publish information on these requirements, including infrastructure gaps, approval timelines and the agencies responsible for addressing them.

It also proposes identifying high-risk or exclusion zones where development should be restricted because of environmental hazards, water stress, land-use conflicts or inadequate infrastructure. States are advised to assess exposure to floods, cyclones, coastal hazards, extreme heat, seismic activity and other environmental risks using forward-looking assessments rather than relying solely on existing hazard maps.

For large data centre campuses, the framework recommends community impact assessments covering traffic, roads, drainage, noise, emissions, waste disposal, emergency access and employment opportunities.

On economic value, the framework raises an important question for state governments: how much does a data centre actually generate in return for the incentives and infrastructure support it receives?

The document recommends distinguishing between passive infrastructure businesses, such as colocation facilities and powered-shell developments, and active computing businesses offering cloud services, GPU-as-a-Service, managed hosting or other computing services. This bifurcation is relevant because the commercial activities, capital expenditure and economic contribution of these facilities may differ significantly.

States should calculate the net economic contribution of an investment after accounting for public support such as land concessions, capital subsidies, electricity-duty exemptions, stamp-duty concessions and infrastructure expenditure, according to the framework. The assessment should also consider local procurement, direct and indirect employment, workforce training, research and development, and the participation of domestic technology suppliers.

The framework recommends linking incentives to measurable outcomes within the control of the organisation receiving them. These may include commissioning milestones, employment commitments, domestic procurement, renewable energy supply and improvements in energy or water efficiency. Clear reporting requirements and conditions under which incentives could be suspended are also proposed.

This approach would allow states to distinguish between announced investment commitments and outcomes that have actually been delivered, according to NITI Aayog.

Save article