In an era where data is considered the new oil and a strategic global currency, data centres act as refineries powering the digital economy. The rise of artificial intelligence (AI) is intensifying the global race to build advanced data and computing infrastructure. As nations scale up digital capacity, integrating sustainability into infrastructure development is becoming essential to ensure long-term responsible growth.

The global AI race is accelerating demand for compute-intensive infrastructure

• Global data boom in the AI-driven digital economy: Global data generation is expanding at an unprecedented pace, with worldwide data volumes projected to reach ~400 zettabytes (ZB)4 by 2028, nearly 2.7 times that of 2024 (IDC Research 2025). This acceleration reflects the rapid digitisation of economies worldwide, driven by rising internet penetration, increased per capita data consumption, the adoption of AI and the growing reliance on digital platforms and services. Among these factors, AI is the primary driver of this massive growth, given its reliance on data-intensive models and continuous learning. In 2025, globally, one out of every six people used generative AI tools to learn, work, or solve problems (Microsoft 2025).
• The US–China AI race is shaping the scale and direction of data centre expansion: Data centres have existed for decades. However, the exponential growth in AI workload has fundamentally transformed data centre intensity. The United States (US) and China lead advanced AI development, accelerating the global race and shaping its pace, scale, and infrastructure.

India is positioning itself as an emerging data centre and AI hub

• India’s growing digital footprint and connectivity expansion: India, with a population of ~1.5 billion (bn), is positioned to play a pivotal role in shaping the global digital evolution, and has been experiencing rapid digitalisation across sectors under the Digital India initiative (UNPF 2025, Digital India 2015). The number of internet subscribers has increased by ~30 per cent over the last 5 years ~0.8 bn in 2020 to ~1.02 bn in 20255 (TRAI 2020, TRAI 2025). The average mobile data consumption in the country reached ~25 GB per user per month in 2024, significantly above the global average of ~19 GB, and is projected to nearly double by 2030 (Ericsson 2024).
• National AI ambitions further reinforcing demand for data centres: India has further strengthened its digital and technological ambitions through the launch of the IndiaAI Mission (PIB 2025), aimed at fostering the development, deployment, and responsible adoption of AI technologies across sectors. India’s approach to AI adoption is intended to be economically grounded and socially responsive (Economic Survey 2026).
• The digital economy is becoming a central pillar of national income growth: The digital economy6 is expected to contribute 20 per cent of national income by 2030, compared to ~12 per cent in FY 23 (MeitY 2025).

Rapid growth intensifies energy, water, and land demands, making sustainability essential

• Electricity demand from data centres is projected to nearly double this decade: Data centres accounted for ~1.5 per cent (IEA 2025) of global electricity consumption in 2024. The demand is projected to nearly double, reaching ~945 TWh by 2030 (equivalent to Japan’s FY24 electricity consumption).
• Water and land use intensify pressure on local communities: Beyond being energy-intensive, data centres also consume large volumes of water and strain land (Gorey 2025), exerting significant pressure on local communities. The global data centre industry currently consumes around 560 bn litres of water annually, equivalent to the domestic water demand of a metropolitan city with roughly 8 million residents (IEA 2025).
• Integrating sustainability and resilience in data centre expansion: Countries, including the US and China, and regions in Europe, have already experienced significant data centre deployment; however, some of these facilities are grappling with water scarcity, energy constraints, and other operational challenges (CERRE 2025). While India’s data centre industry is still nascent, it is poised for a rapid scale-up in the coming years. Given India’s already critical resource constraints (Mehra 2025, World Bank 2025), and its large population, sustainability and resilience must be integrated from the outset to support long-term growth while minimising environmental and social impacts.

Our approach

• Understanding the data centre landscape: This study assesses the current state of the sector by examining market dynamics, the policy and regulatory environment, energy procurement strategies, and industry practices. While doing so, it identifies opportunities to embed resilience and sustainable practices as a core design and operational principle, and highlights priority research areas that must be addressed as the sector scales.
• Combining literature with India-specific stakeholder insights: This study adopts a hybrid framework that integrates a review of the existing literature with primary insights from stakeholder consultations (one-on-one interviews and survey responses). A total of 24 engagements were conducted with key stakeholders across the data centre ecosystem, including data centre operators, cloud service providers, renewable energy (RE) developers, policymakers, and domain experts.

Power and water availability, along with the climate risk profile, are the primary parameters considered when selecting a site for a data centre. Beyond these factors, several additional operational and logistical parameters must also be considered (Figure 3). Given that data centres are designed for long operational lifecycles, it is essential to carefully evaluate all these parameters to ensure operational reliability, cost efficiency, and longterm performance.

Figure 3: Critical factors that define data centre site selection

Access to reliable power is the primary siting criterion

• Continuous, uninterrupted power is fundamental to data centre operations: Operating a data centre requires a stable, uninterrupted 24/7 power supply. Servers, storage systems, cooling units, and other network equipment must operate simultaneously to maintain data operation. Server and cooling infrastructure collectively account for nearly 70 per cent of the total electricity consumption in data centres, though this proportion varies widely by facility type (IEA 2025). Even a momentary power loss can cause servers to shut down abruptly, leading to data corruption, disrupting any real-time operation that relies on the particular data centre, and causing a widespread service outage.
• Power disruptions directly translate into contractual and financial risk: Typically, uptime requirements16 are contractually defined between the data centre operator and the end user (tenant), with any power disruption or service outage constituting a breach of service-level agreements (SLAs) and resulting in financial penalties for the operator. A 2024 global survey on data centres indicates that 54 per cent of data centres reported annual outage costs exceeding USD0.1 million, with ~20 per cent of operators reporting costs exceeding USD1 million (Uptime 2024). Outage costs extend beyond immediate losses to contracts, customer retention, and insurance.

Water availability influences data centre planning

• Cooling needs make water availability a critical siting consideration: As servers process vast volumes of data, they generate substantial heat, requiring effective cooling systems to prevent overheating, equipment failure, or permanent damage (Annexure). Data centres relying on waterbased heat-rejection systems17 , such as cooling towers, have high overall water consumption (Mytton 2021). The volume of water required depends on the data centre’s size, IT load, local climate, and cooling technology used.
• Water unavailability may cause operational challenges: Given that water is an increasingly constrained resource, careful assessment of water availability and sustainability is essential in data centre planning. Locating data centres in already water-stressed areas can place significant pressure on local communities (CERES 2025, Triona McGrath 2024, World Economic Forum 2024) and may affect the long-term operational viability of the facilities. Yet, global trends show a growing imbalance, with around 43 per cent (S&P Global 2025) of global data centres operating in areas of high water stress in the current decade.
• Rising rack densities further intensifying water use: Driven by AI and high-performance computing workloads, average rack power densities have increased from approximately 1–3 kW (Eisenband 2023) in the early 2000s to around 8 kW (Uptime 2024) per rack today. Rising rack densities increase cooling intensity, and in facilities that are reliant on evaporative cooling, it can intensify water use unless offset by liquid or closed-loop cooling technologies. A typical 100-MW hyperscaler data centre can consume ~2 million litres of water per day for on-site cooling (IEA 2025).

Climate risks increasingly affect operational resilience

• Climate risk profile must be prioritised in data centre planning: The climate risk profile, including exposure to flooding, seismic activity, soil movement, tropical cyclones, wind, and coastal inundation, of potential sites needs to be factored in. If these profiles are not assessed during installation, and mechanisms to mitigate these challenges are not established, data centres can suffer significant damage (Shine Jacob 2023) and financial losses. Studies suggest that globally, ~6 per cent (XDI 2025) of installed data centres are located in zones highly exposed to climate-related hazards, while 16 per cent are in moderate-risk zones.
• India’s climate vulnerability amplifies siting risks: As many as 72 per cent (Wadhawan 2023) of Indian districts are exposed to extreme flood events, while 57 per cent (Shravan Prabhu 2025) face high to very high risk from extreme heat. Exposure to extreme heat can significantly increase cooling costs, while flooding poses direct risks to physical infrastructure. This underscores the need for a detailed assessment before setting up a data centre. While India’s data centre industry is still in its nascent stage, such assessments of environmental, climate, and infrastructure risks should be prioritised to ensure sustainable, resilient, and long-term operations as the sector scales. This perspective is reinforced by our stakeholder consultations, which indicate that operators are increasingly factoring climate risk profiles into data centre planning and operational decisions.

In addition to the aforementioned parameters, land availability, the policy and regulatory landscape, and permitting and approval processes frequently determine project viability and development timelines. Connectivity-related factors such as access to optical fibre networks, proximity to subsea cable landing stations, and latency requirements are also critical to ensuring high-speed, low-latency operations. The relative importance of these parameters, however, varies significantly depending on the intended function of the facility, whether hyperscale, colocation, or edge (refer to the Annexure).

India does not yet have a national policy framework specifically governing data centre development. A Draft National Data Centre Policy (MeitY 2020) circulated by the Ministry of Electronics and Information Technology (MeitY) in 2020 was not notified; instead, central support has been provided through enabling measures such as according infrastructure status (The Hindu 2025) to data centres above 5 MW in 2022 and the announcement of a long-term tax holiday (PIB 2026) for data centre and cloud investments (extending up to 2047) in the Union Budget 2026. In the absence of a binding national framework and with national guidance largely limited to an early compilation of energy-efficiency best practices, (BEE 2010), state governments have taken the lead. Currently, 15 states support data centres through dedicated policies or through IT/ITeS and industrial policies (see Figure 4).

Table 1. Comprehensive comparative matrix of the policy levers across states

Source: Authors’ compilation from 15 states’ policies

State-level policies in India prioritise investment attraction over performance outcomes

• Incentives have become the primary basis of state competition: Stamp duty exemptions, electricity duty waivers, capital subsidies, and expedited approvals have effectively become standard policy tools, with states competing primarily on the scale, duration, and bundling of incentives.
• Power-related incentives dominate state policy design: States such as Uttar Pradesh, Rajasthan, Odisha, Gujarat, Haryana, and Tamil Nadu place central emphasis on power concessions, including tariff rebates, electricity duty waivers, transmission and wheeling charge exemptions, and facilitation of captive and open-access procurement. Consultations indicate these measures are critical for lowering delivered electricity costs, the single most important determinant of data centre viability.
• Data centre park models reflect hyperscaler preferences for speed and certainty: Gujarat, Rajasthan, Uttar Pradesh, Haryana, and Andhra Pradesh have adopted park-based development models that offer plug-and-play infrastructure, pre-zoned land, floor-area ratio (FAR) relaxations, bundled approvals, and common utilities. This aligns with hyperscaler priorities for scale-ready, timebound deployments.
• A subset of states is embedding explicit sustainability objectives: Rajasthan, Odisha, Karnataka, Tamil Nadu, and Andhra Pradesh have begun integrating sustainability into policy design. Notably, Rajasthan’s 2025 policy constitutes India’s most comprehensive sustainability-oriented data centre framework, integrating measures such as mandatory wastewater recycling, zero liquid discharge (ZLD), rainwater harvesting, groundwater recharge, advanced water management systems, green building standards, and targeted subsidies for green solutions. This integrated approach positions Rajasthan as a reference model for climatealigned data centre expansion in water-stressed geographies.
• States are innovating on sustainability tools: Andhra Pradesh promotes seawater cooling for coastal data centres, leveraging natural geography to reduce cooling-related electricity demand. Karnataka and Tamil Nadu emphasise green building certification and energy-efficient infrastructure, while Odisha explicitly links data centre development to carbon footprint reduction.
• Three systemic gaps persist across most state policies: Across most states, data centre policies exhibit three recurring gaps—(i) a water governance deficit, with limited provisions on cooling water management and reuse; (ii) an absence of performance standards such as power usage effectiveness (PUE), water usage effectiveness (WUE), or carbon benchmarks; and (iii) weak climate resilience framing, despite growing exposure to floods, cyclones, heatwaves, and seismic risks. In practice, some states are already implicitly accounting for water constraints in siting decisions. For instance, Maharashtra has prioritised coastal and peri-urban clusters, such as Navi Mumbai’s Rabale hub, while slowing development in waterstressed inland areas, such as Nagpur. Yet these considerations are not codified through enforceable water-efficiency norms or policy requirements.

As hyperscale capacity expands, future policy leadership will depend less on fiscal generosity and more on systemic integration of energy, water, land, and climate resilience. Rajasthan, Odisha, Karnataka, Tamil Nadu, and Andhra Pradesh offer early templates for such nextgeneration policy frameworks.

Global policy approaches increasingly link incentives to performance

• Regulation is shifting from permitting to managing system-wide impacts: Globally, governments are increasingly tightening regulatory oversight of data centres as sustainability ambitions rise and electricity systems come under growing stress. Policy focus is shifting beyond site-level permitting towards managing cumulative system impacts through a combination of mandatory reporting, efficiency standards, renewable integration requirements, and grid-connection conditions.
• The European Union leads on performancebased regulation: The revised EU Energy Efficiency Directive (2023) (EUDCA 2025) mandates energy and water reporting for data centres above 500 kW, and requires waste heat recovery for facilities above 1 MW. The directive also signals expectations for greater renewable energy uptake and continuous efficiency improvements. These obligations are complemented by voluntary initiatives such as the European Code of Conduct for Energy Efficiency in Data Centres, (EU 2025) and emerging provisions under the EU AI Act, which increasingly link digital infrastructure development with energy-system performance. Several EU countries are going further. Germany’s Energy Efficiency Act (EnEfG) (EnEfG 2023) imposes binding requirements on efficiency, renewable sourcing, waste-heat utilisation, and energy management systems. At the same time, Ireland has introduced stringent grid-connection rules for large energy users, requiring on-site generation and storage to limit system stress. In parallel with the Energy Efficiency Directive, the EU is also developing a harmonised data centre rating scheme, (EU 2026) expected to be introduced from 2026 onward.
• Asian economies are linking approvals to sustainability metrics: China’s Green Data Centre Standards (Shid 2025) sets caps on water-to-energy ratios for cooling. Singapore’s Green Data Centre Roadmap (IMDA 2024) and Green Mark for Data Centres (BCA 2024) establish rigorous standards for energy performance, monitoring, and reporting. Malaysia is similarly advancing with the introduction of national guidelines (Skrine 2025) on power, water, and carbon efficiency, which are increasingly linked to project approvals and associated incentives.
• The United States follows a fragmented, stateled approach: In contrast, regulation in the US remains fragmented, with limited federal mandates and greater reliance on state policies, utility interconnection rules, and local permitting processes. While federal efforts emphasise voluntary efficiency programmes, several states are introducing harder constraints. Virginia (Virginia Legislative 2024) links tax exemptions to meeting efficiency and clean power criteria, while Texas (Texas Legislature 2025) is tightening connection rules for huge loads, including curtailment obligations during grid emergencies, reflecting rising concerns about grid reliability. Arizona’s Tucson city (Mendoza 2025) mandates conservation plans for large water users and imposes penalties for exceeding thresholds.

Compared to these jurisdictions, India’s regulatory approach remains largely incentive-driven and nonprescriptive, relying on state-level industrial policies, general environmental compliance, and renewable procurement mechanisms. This highlights a growing gap between India’s investment-led data centre expansion and global shifts towards performance-based sustainability regulation.

Stakeholders highlight implementation gaps despite supportive policy intent

• Single window in policy, delays in practice: Industry stakeholders consistently flagged slow, fragmented approvals as a material driver of project delays (particularly grid connectivity approvals, building plan sanctions, and fire NOCs). This suggests that, despite many state policies offering a single-window mechanism, the delays persist in practice. Our consultations further suggest that, after power, land acquisition is the most significant hurdle at present, while the perceived water-availability risk remains low. However, this likely reflects the sector’s early stage and may understate future constraints as deployments scale up. This reinforces the need for proactive planning now.
• Regulatory friction in clean energy procurement is raising project risk and costs: Consultations indicate that institutional and regulatory barriers now outweigh technological constraints. Key institutional barriers include protracted approval processes, which often delay renewable sourcing decisions or lead to project deferrals, despite the presence of deemed-approval provisions. Data centre operators and renewable energy developers independently highlighted high cross-subsidy surcharge (CSS) and additional surcharge (AS), inconsistent banking rules, sudden withdrawal of waivers/provisions, and state-wise policy heterogeneity as the principal regulatory impediments. Collectively, these factors elevate project risk premiums, undermine long-term contracting, and increase financing costs.
• Integrated land–energy–water planning is seen as essential for sustainable siting: Domain experts strongly endorsed the development of integrated national spatial planning frameworks to guide data centre location decisions, rather than fragmented industrial zoning approaches. Sustainable data centre expansion, they argued, requires co-optimisation of grid readiness, cooling water availability, land suitability, and climate resilience. Urban-edge projects like Actis’s green data centre in Taipei (SYSTEMIQ 2025) showcase how integrated land, water, and energy planning can enable sustainability even in dense, constrained environments. In our consultations with data centre operators as well, a strong consensus emerged that India requires dedicated green data centre policy frameworks, rather than relying solely on generic industrial promotion policies.