Initialising Intelligence
27 November 2026 | Gandhinagar, Gujarat
Hosted by Department of Science & Technology, Government of Gujarat
Vibrant Gujarat Global Summit, 10-12 January 2027
Hosted by
Department of Science & Technology, Government of Gujarat Government of Gujarat Gujarat Informatics Limited
An Initiative by Elets Technomedia
Knowledge Partner eGov Magazine
  1. Home
  2. Blogs
  3. AI Infrastructure
AI Infrastructure

AI Data Centres in India: Infrastructure, Investment and Sustainability Challenges

AI data centres in India have moved from a niche real estate segment to a national strategic priority. The opportunity is large. So are the questions about power, water and who pays the environmental bill.

EG
eGov Editorial TeamElets Technomedia
6 min readAI Infrastructure
AI data centres in India highlighting digital infrastructure, investment, power requirements and sustainable cooling technologies
Every AI reply runs on a building somewhere. India is racing to build more of them at home.

Every chatbot reply, every model training run and every AI-powered public service ultimately runs on physical machines in a building somewhere. As AI demand surges, AI data centres in India have moved from a niche real estate segment to a national strategic priority. The opportunity is large. So are the questions about power, water and who pays the environmental bill.

Key takeaways

  • India generates nearly 20% of the world’s data but hosts only 3–4% of global data centre capacity.
  • Operational capacity grew from about 375 MW in 2020 to nearly 1.8 GW in H1 2026, with a pipeline of nearly $90 billion.
  • Electricity and water are the binding constraints; data centre power demand could reach 13.56 GW by 2031–32.
  • Renewable power, efficient cooling and transparent reporting are the keys to building responsibly.

The Scale of the Buildout

India generates a huge share of the world’s data but hosts only a small share of its computing capacity. The country generates nearly 20% of the world’s data yet accounts for only 3–4% of global installed data centre capacity. That mismatch is the core of the investment story.

375 MWOperational capacity in 2020
~1.8 GWOperational capacity in H1 2026
890 MWIn Mumbai, half of the total
~1.7 GWMumbai’s upcoming pipeline

Capacity has already grown fast. Operational capacity rose from approximately 375 megawatts in 2020 to nearly 1.8 gigawatts in the first half of 2026. Geography is concentrated: Mumbai accounts for half of operational capacity at 890 MW and has the largest upcoming pipeline at around 1.7 GW, though expansion is spreading to other hubs.

Forecasts for 2030 vary widely depending on assumptions. The Economic Survey 2025-26 projected capacity more than tripled to about 4 GW, while Wood Mackenzie projects 12 GW, with AI-dedicated capacity growing nearly 24-fold from 275 MW to 6,546 MW. Treat these as a range rather than a prediction. The consistent message is steep growth, with AI as the main driver.

Where the Money Is Going

Investment is following demand. The sector’s development pipeline has swelled to nearly $90 billion, fuelled by AI, cloud and digital services. Some of the headline projects are very large:

  • Google, Visakhapatnam. An AI and data centre hub as part of a five-year, $15 billion investment from 2026 to 2030, developed with AdaniConneX and Airtel.
  • Reliance, Jamnagar. Reliance Industries is building a 3 GW AI-ready data centre in Jamnagar, Gujarat.
  • CapitaLand India Trust. More than ₹5,000 crore committed to data centre developments in Navi Mumbai, Hyderabad and Chennai.

This is the physical foundation of AI infrastructure: not just servers, but land, grid connections, cooling systems, fibre networks and skilled operators. It also supports India’s sovereign AI ambitions by keeping sensitive data and critical workloads within national borders.

Related: The Rise of Sovereign AI in India: Why Infrastructure, Data and Innovation Matter

The Power Problem

Electricity is the most immediate constraint. AI workloads are far more power-intensive than traditional computing. The Economic Survey warned that data centres need large volumes of electricity and water and that AI workloads introduce volatility into power demand, risking grid stability.

The scale of future demand is significant. The Ministry of Power has estimated that electricity demand from data centres could reach 13.56 GW by 2031-32. Because data centre load clusters in a few states, local grids can feel the strain well before national supply does.

That makes clean energy central to the plan. Operators are increasingly looking at captive solar and wind, hybrid power purchase agreements, battery storage and open-access power. The goal is to avoid locking in a large new fossil-fuel footprint just as the sector scales.

The Water Question

Water is the less visible but potentially sharper challenge. Cooling dense AI hardware takes a lot of it in a country where many regions already face scarcity. The Economic Survey said power, finance and especially water remain limited.

There is also an indirect cost. AI’s electricity demand draws on power plants that themselves need large volumes of water, and this off-site footprint can exceed what the data centre uses directly. Transparency is lacking: data centre regulation is largely left to state governments, and only five of the 15 states with data centre policies mention sustainability-related issues in their guidelines.

Better cooling technology can help. Closed-loop and liquid cooling reduce consumption, and sustainability was a central theme at industry gatherings such as the AI-Powered Data Centre Conference in Mumbai in July 2026, where speakers described AI as a compute story, a power story, a cooling story, a water story and a data-governance story.

AI is a compute story, a power story, a cooling story, a water story and a data-governance story.Speakers at the AI-Powered Data Centre Conference, Mumbai, July 2026

Other Hurdles to Watch

  • Hardware access. The Economic Survey also flagged GPU constraints. Advanced chips remain costly and dependent on global supply chains.
  • Execution risk. Announcements are easy; commissioning is hard. Land, approvals, grid connections and skilled labour often decide timelines.
  • Uneven policy. State-level incentives vary, and clear national standards on efficiency and water reporting are still developing.
  • Financing and tariffs. Competitive power pricing and reliable captive generation will shape which projects are viable.

Building Responsibly

The path forward combines ambition with discipline. Practical priorities include:

  1. Pairing every major campus with credible renewable power.
  2. Adopting water-efficient or closed-loop cooling by default.
  3. Publishing energy and water metrics so communities and regulators can hold operators accountable.
  4. Locating facilities with grid and water capacity in mind, not only tax incentives.
  5. Setting consistent sustainability expectations across states.

Why Convening Matters

No single player can solve these issues alone. An AI technology summit brings together data centre developers, utilities, chip makers, policymakers and enterprise users to negotiate trade-offs openly: how fast to build, how to power it and who bears the costs. For operators and investors, these forums are also where partnerships form and where emerging rules get tested before they are finalised.

The Road Ahead

India’s data centre story is one of rare momentum: rising demand, large capital commitments and a clear national interest in hosting AI at home. The test will be whether the buildout stays resilient and sustainable as it scales.

If the sector solves the power and water equation, AI data centres in India could become a lasting competitive advantage. If it does not, the same boom risks straining grids and local water supplies. The next few years of planning, regulation and engineering choices will decide which story gets told.

Where the Conversation Continues: Gujarat AI Summit 2026

Building AI data centres at scale is about more than adding servers. It means getting digital infrastructure, power, data security and governance right from the start, and that takes honest conversation between the people planning, funding and regulating these projects.

The Gujarat AI Summit 2026 is designed for that kind of dialogue. Its agenda covers AI governance, digital infrastructure, data security, research and responsible AI, the themes that will decide whether India’s AI infrastructure grows in a way that is secure, sustainable and built to last.

If you are developing facilities, investing in the sector, shaping policy or planning how your organisation will use AI compute, the summit is a chance to compare approaches, hear what is working on the ground and connect with the people building India’s digital backbone.

Connect with the people building India’s digital backbone

Compare approaches with developers, utilities, investors and policymakers at Gujarat AI Summit 2026 in Gandhinagar.

Register for the Gujarat AI Summit 2026
Read nextIndia’s AI Innovation Ecosystem: Startups, Research and the Future of Technology
EG

eGov Editorial Team

The editorial desk of eGov magazine by Elets Technomedia, covering e-governance, smart cities and public-sector digital transformation across India.

Link copied
Our Publications

Elets Magazines — Latest Issues

Five sector magazines with interviews, case studies and data from the people running India’s biggest programmes.