Microsoft Opens Fourth Indian Cloud Region as Data-Centre Capacity Must Grow Five- to Seven-Fold

The news

New Delhi / Hyderabad. Microsoft on Monday announced the launch of its fourth cloud region in India — India South Central, in Hyderabad — which it is positioning as a strategic hub for Asia and the Global South, The Economic Times reports. Its other Indian cloud regions are in Pune, Chennai and Mumbai. A cloud or data-centre region is a location where a company builds a cluster of data facilities that store and process customers’ data and run software, including artificial intelligence models. Puneet Chandok, Microsoft’s president for India and South Asia, told ET on the sidelines of the launch event that the company is constantly scouting for new regions as demand for cloud and AI infrastructure rises, because “these regions take a lot of time to build”. He said India currently has about 2 GW of data-centre capacity but will need 10-14 GW in the coming years. Microsoft has committed $20.5 billion towards cloud and AI infrastructure in India, comprising $17.5 billion announced in December 2025 on top of an earlier $3 billion commitment. Globally, the company is expanding its data-centre capacity from around 12 GW to more than 38 GW by 2032; Mr. Chandok did not disclose India’s share but said the country would remain a “meaningful part” of it. He said building AI infrastructure would need hyperscalers — the very large cloud providers — as well as private capital, government support and smaller cloud providers, and that customers increasingly want infrastructure that gives them greater control over their data. “We do not train our foundation models on commercial customer data,” he said, adding that AI “and agents that we’re building always need to be in human control”. Jitin Prasada, Minister of State for Commerce and Industry and for Electronics and Information Technology, said at the same event that there is “no compromise when it comes to the environmental impact or excessive use of water by data centres”, and that the government is working on ease of doing business and land availability as capacity expands. The same day, ET reported a Wall Street Journal analysis that major technology companies hold $3.1 trillion in off-balance-sheet commitments, with Microsoft’s uncommenced leases, primarily for data centres, at $329.1 billion. The syllabus link is infrastructure, investment models and digital sovereignty: data centres combine land, electricity, water and capital into the physical base of the digital economy.

Static syllabus linkage

1. Data centres are now recognised as infrastructure. In October 2022 the Union Finance Ministry added data centres to the Harmonised Master List of Infrastructure sub-sectors, which makes such projects eligible for easier and longer-term infrastructure lending. The Ministry of Electronics and Information Technology released a draft Data Centre Policy in 2020 to promote India as a data-centre hub, and several States, including Telangana, Maharashtra, Tamil Nadu and Uttar Pradesh, have their own data-centre policies offering land, power and tax incentives. For government workloads, MeitY empanels cloud service providers under the MeghRaj or GI Cloud initiative, launched in 2014.
2. Where data sits is governed by several laws. The Digital Personal Data Protection Act, 2023 allows transfer of personal data outside India except to countries that the Central Government restricts by notification, a “negative list” approach, and establishes the Data Protection Board of India for enforcement. Sectoral rules can be stricter: the Reserve Bank of India’s directive of April 2018 requires payment system data to be stored only in India. Such localisation rules are one reason global cloud firms build regions inside India, because regulated customers such as banks and government departments need data to stay within the country.
3. Compute is a pillar of India’s AI strategy. The IndiaAI Mission, approved in March 2024 with an outlay of about ₹10,372 crore, includes a compute pillar under which the government subsidises access to graphics processing units for researchers and start-ups through private providers. Large language models require clusters of such processors, which consume large amounts of electricity and need cooling, often using water. This links digital policy to energy and water policy, making data centres a cross-sectoral concern.
4. Power and water are the physical limits of the digital economy. A data centre’s capacity is measured in megawatts or gigawatts of electrical load, not in floor area, because power is the binding constraint. Cooling can use evaporative systems that consume water, and efficiency is often expressed as Power Usage Effectiveness, the ratio of total facility energy to the energy used by computing equipment. India’s target of 500 GW of non-fossil electricity capacity by 2030 and its water stress in many cities mean that data-centre growth must be planned with the grid and water resources in mind.

Why UPSC loves this

1. Infrastructure and investment models are core GS3. The GS3 syllabus covers infrastructure, investment models and the effects of liberalisation on the economy. UPSC has asked about data localisation, the digital economy and cloud computing in earlier papers. Data centres are the meeting point of these topics and are increasingly discussed in Economic Surveys.
2. Digital sovereignty is an emerging essay and GS2 theme. Questions on data protection, data localisation and the power of big technology firms have appeared in both GS2 and essays. The Microsoft story adds the physical dimension: sovereignty depends not only on laws about data but on who owns the buildings, the chips and the power supply.

Analysis

1. The gap between 2 GW and 14 GW is really a power-sector problem. Going from about 2 GW to 10-14 GW of data-centre load means adding demand equivalent to several large power stations, concentrated in a few cities and required round the clock. Firms want this power to be green to meet their climate commitments, which requires either dedicated renewable capacity with storage or open-access procurement through State grids. The binding constraint is therefore the capacity of State distribution companies and transmission networks, not the willingness of hyperscalers to invest. Data-centre policy that ignores the grid will produce announcements rather than capacity.
2. Foreign capital builds the base, but dependence has a price. Microsoft’s $20.5 billion commitment brings capital India does not have to spare, and in the language of factors of production it adds physical capital and uses land and energy productively. But when the largest facilities are owned by a handful of foreign firms, pricing power, access to advanced chips and decisions about where AI capacity is placed rest outside India. The counter-view is that ownership matters less than jurisdiction, because data centres located in India are subject to Indian law. The balanced policy is to welcome hyperscalers while supporting Indian providers and public compute so that the market does not become a closed oligopoly.
3. The Minister’s water promise needs a mechanism. Saying there will be “no compromise” on water use is easy; enforcing it requires water-use disclosure, efficiency standards and siting rules that keep water-intensive cooling out of stressed basins. Hyderabad, Chennai and Bengaluru have all faced water shortages in recent years. A data-centre policy could require reporting of water use per unit of computing, favour air or closed-loop cooling in water-stressed cities and link approvals to local water budgets. Without such rules, the promise will not bind anyone.
4. Off-balance-sheet commitments signal a financial risk worth watching. The $3.1 trillion in off-balance-sheet commitments reported by the Wall Street Journal shows that the AI build-out is being financed partly through long leases and future obligations rather than current cash. If demand for AI services grows more slowly than expected, some projects could be delayed or abandoned, leaving land and power contracts stranded. For India, this argues for sequencing public support — land allotment, grid connection — to actual construction milestones rather than announcements. Enthusiasm for investment should not blind policymakers to the cycle.
5. Customer control of data is becoming a competitive promise. Mr. Chandok’s assurance that Microsoft does not train foundation models on commercial customer data shows that trust has become a selling point. For Indian regulators, this is an opportunity to turn voluntary promises into enforceable standards through contracts for government cloud services and rules under the data protection law. Businesses and government departments need assurance not only about where data is stored but about what it is used for. The shift from localisation to use-limitation is the next stage of data governance.

Administrator's brainstorm