Crisil sees India's data centre capacity growing five-fold to 7.5 GW by 2030, drawing over ₹3 lakh crore

Crisil Ratings expects a substantial expansion in Indian data-centre capacity over the next four years. In a report released on 3 November 2026, the agency projected that installed data centre capacity will grow about five-fold to around 7.5 gigawatts (GW) by March 2030, drawing more than ₹3 lakh crore of investment over the four financial years to fiscal 2030. Crisil described the sector as entering an "investment super-cycle".

The forecast matters for India's deep-tech ambitions because data centres are the physical base under almost everything else: training and serving AI models, running government digital platforms, and keeping sensitive data inside the country. How quickly that base grows, and where, will shape how far India's AI build-out can go.

The headline numbers

The report’s projections concern third-party colocation data centres. Installed capacity includes completion of mechanical, electrical and plumbing work; it is not a measure of AI training compute already in service. Its main projections are:

• Capacity: about 7.5 GW by March 2030, roughly five times today's installed base.
• Investment: more than ₹3 lakh crore over the four years through fiscal 2030.
• Pipeline: of the roughly 6 GW still to be built, about three-fourths is expected in Visakhapatnam, Mumbai and Hyderabad.
• Demand visibility: nearly half of the upcoming capacity is already tied up in long-term contracts. More than 95% of the capacity now under construction is pre-contracted.
• Utilisation: existing facilities are running at about 90%.
• Hyperscaler share: large cloud providers ("hyperscalers") are expected to lease 75–80% of industry capacity by fiscal 2030, up from 60–65% in fiscal 2026.
• Cost edge: development costs in India are about one-fifth lower than in several competing Asia-Pacific markets. Related coverage says lower costs and power tariffs make Indian data centre rentals about 30–35% cheaper.

Why the build-out is speeding up

Several forces are adding up at once.

AI workloads. Training and running large AI models needs dense racks of GPUs that draw far more power and produce far more heat than ordinary servers. That is pushing operators toward bigger campuses, higher power density and liquid cooling. Meeting higher rack densities requires investment in suitable power and cooling infrastructure.

The geography is shifting

Mumbai has long been India's data centre capital, thanks to its submarine cable landings and financial-sector demand. Crisil's projection that Visakhapatnam, together with Mumbai and Hyderabad, will host about three-quarters of the new capacity points to a real change. Visakhapatnam, which until recently was barely on the data centre map, is now expected to be one of its three biggest growth centres.

Chennai and the Delhi-NCR region remain important too, as the recent run of announcements there shows. The broader point is that capacity is moving to wherever large blocks of land, reliable power and connectivity can be secured together.

The constraints

Crisil was clear that the pace of expansion depends on a few things going right.

Power. A 7.5 GW fleet needs a lot of steady electricity. Timely grid connections are the single biggest dependency, and operators increasingly want that power to be renewable to meet their customers' climate targets. That links the data centre boom directly to India's work on transmission, battery storage and round-the-clock renewable supply.

Approvals. Land, environmental and power-connection approvals can each delay a project by months.

Execution and pricing. As new capacity comes online quickly, Crisil warned that pressure on rentals could weaken project returns. High utilisation today does not guarantee the same once supply catches up.

What it means for India's deep-tech stack

A bigger, cheaper domestic data centre base helps India's AI ecosystem in practical ways. Start-ups building Indic-language models, AI inference platforms and enterprise AI tools need affordable compute close to their users. Data protection rules and government preferences for keeping sensitive data in-country push the same way.

It also creates demand further down the supply chain: power electronics, cooling systems, high-speed networking and, over time, chips designed and packaged in India. Whether Indian suppliers capture that demand, or whether most of the hardware is imported, is one of the bigger questions behind the headline investment figure.

Crisil's forecast doesn't guarantee the build-out. But with half the coming capacity already contracted and current sites nearly full, the demand signal is stronger than in most infrastructure booms. Power and execution will decide how quickly it turns into working capacity.

Sources

• https://www.business-standard.com/technology/artificial-intelligence/india-s-data-centre-capacity-expected-to-increase-five-fold-by-2030-126110300990_1.html
• https://www.business-standard.com/amp/technology/tech-news/india-s-data-centre-sector-to-draw-3-trillion-investment-by-fy30-crisil-126110301259_1.html
• https://www.deccanherald.com/business/india-set-for-five-times-surge-in-its-data-centre-capacity-by-2030-4205631
• https://www.equitybulls.com/news/378461/crisil-ratings-india-s-data-centre-capacity-set-for-5x-surge-by-2030
• https://www.saurenergy.com/solar-energy-news/lower-costs-tariffs-make-india-data-centre-rentals-30-35-cheaper-crisil-12779280
• https://news.microsoft.com/source/asia/features/microsofts-newest-india-datacenter-region-goes-live-to-power-the-countrys-ai-economy-and-enable-frontier-firms/
• https://www.adani.com/newsroom/media-releases/adani-commits-usd-100-bn-to-sovereign-ai-infrastructure
• https://www.kotakneo.com/news/stocks/adani-group-seeks-funding-for-100-billion-digital-infrastructure-plan/
• https://www.medianama.com/2026/10/223-meity-ai-compute-providers-gpus-indiaai-mission/
• https://www.crisilratings.com/en/home/newsroom/press-releases/2026/11/indias-data-centre-capacity-set-for-5x-surge-by-2030.html