The government has moved to dispel concerns about water supply impacts from data centres, with a senior official stating that the belief that growing data centres will automatically drain local water supplies in India is flawed. According to the official, this concern is based on outdated, evaporative cooling technologies used in older global facilities. Modern data centres use advanced closed-loop, air-cooling and direct-to-chip liquid-cooling systems that continuously recirculate water internally rather than evaporating it into the atmosphere, reducing daily operational water demand to near zero. Efforts are underway to make all new data centres operate on these systems, with the official emphasizing that the government is working to ensure that freshwater consumption for such upcoming facilities remains negligible.
Yotta Data Services CEO Sunil Gupta addressed growing concerns about data centre infrastructure strain at Global Fintech Fest 2026, stating that large operators can build dedicated power networks and closed-loop cooling systems to limit dependence on local infrastructure. Gupta explained that India's data centre industry began building facilities of 50 MW or more only around 2016-17, by which time operators understood that securing continuous fresh water supply would be difficult. Most modern Indian data centres use air cooled chillers with closed-loop systems that circulate water internally without continuous fresh water consumption. Yotta is setting up a 80 MW building in Mumbai and plans to build a 120 MW facility in Greater Noida to accommodate expansion, with more than 85% of demand expected from overseas customers. The larger challenge is the ability of local distribution networks to carry power to large facilities, with data centres in urban areas potentially straining local power infrastructure.
Bharti Airtel has advocated for a comprehensive sovereign technology stack approach to control Indian data processed overseas, according to Group CRO and Director -- Corporate Affairs Rahul Vatts speaking at Global Fintech Fest 2026. At the panel titled "Building Sovereign AI for a Digital Economy," Vatts emphasized that data residency alone does not amount to sovereignty - describing the term as having been reduced to a checkbox exercise where keeping data within India's borders is treated as sufficient on its own. The company outlined four principles that define sovereignty in practice, with others in the industry increasingly adding a fifth principle of technological sovereignty - covering control over the underlying technology stack itself. Vatts stated that true sovereignty is a systemic property of the entire AI stack rather than a location attribute of a dataset, with the distinction being significant for AI infrastructure where workloads can be hosted in Indian data centres while parts of the software, management layer, operational processes or technology stack remain subject to external control.
According to Vatts, the government's first formal move came through a circular issued on March 20, 2026, followed by roughly a month to six weeks of detailed discussion on implementation. The company official highlighted that two areas should always be important for us to keep - clear discussion on where the control plane is and the manner in which it will be managed. In the Union Budget for 2025-26, Finance Minister Nirmala Sitharaman had announced a tax break until 2047 for companies setting up data centres, which is likely to make India an attractive destination for new data centres. The policy push is already showing results, with India adding 387 megawatts of data centre capacity in 2025 alone, according to a report from Rubix Data Sciences. India ranked seventh globally by number of data centre facilities as of June 2025, with Mumbai, Hyderabad, Delhi-National Capital Region, Bengaluru, and Chennai together accounting for nearly 65 per cent of the country's total data centre capacity.
India is already generating the largest data per consumer globally, with 38-40 GB per month per customer currently and predictions suggesting 70-75 GB per customer in next 3-4 years, as reported by Vatts. The country is experiencing more than 2.4 billion UPI transactions and increasingly digitizing all systems. Vatts noted that today your data is susceptible to anything - including identity, health records, and financial reports - and questioned why any other country should have access to that data. The scale of India's digital economy adds urgency to the sovereignty debate, with AI is likely to increase that dependency further as training and inference workloads require large amounts of compute, storage, networking and data movement. As AI becomes embedded into enterprise and public-sector systems, more sensitive datasets will increasingly interact with infrastructure layers, meaning sovereignty cannot be isolated to the database layer - it has to extend across the infrastructure supporting data collection, processing, model execution and application delivery.
According to Vatts, the world is also trying to look at it differently due to the geopolitical environment, with countries like Germany and France increasingly needing to control their data. As sectoral regulators begin issuing enabling guidelines, the question for critical workloads is no longer just about cloud selection but careful evaluation of which cloud architecture guarantees end-to-end control under Indian law. The sovereign AI discussion also brings telecom infrastructure into the cloud conversation, with AI workloads increasingly depending on high-capacity networks connecting data centres, enterprises, edge locations and users. Telecom operators with domestic fibre networks, data centres, points of presence and cloud capabilities are consequently positioned across multiple layers of the infrastructure stack. For critical sectors, the eventual definition of sovereign cloud is likely to depend on several interconnected layers including physical location, data governance, control planes, operational access, legal jurisdiction and technology dependencies, making sovereignty less a cloud procurement decision and more an infrastructure strategy. The sector's investment pipeline stands at around $90 billion, nearly six times the $13-15 billion invested between 2020 and 2024, with AWS, Microsoft, and Google together committing approximately $67.5 billion to India.
