India's Data-Centre Capacity Passes 1.57 GW - the Binding Constraints Are Now Power, Land and Water
- India's data-centre capacity rose from about 375 MW (2020) to 1.57 GW (Aug 2026), with the government projecting nearly 8 GW by 2030 against roughly 17 GW of demand by 2031-32.
- About USD 70 billion has been invested with a further USD 90 billion announced; data centres have held infrastructure status since the 2022-23 Budget.
- The 2026-27 Budget adds a tax holiday for eligible foreign cloud service providers through 2047, aimed at pulling hyperscale capacity onshore.
- The binding constraints are now power, land and water - plus the imported silicon inside - not demand; the gap between announced and built capacity is the execution test.
- India's data-centre capacity rose from about 375 MW (2020) to 1.57 GW (Aug 2026), with the government projecting nearly 8 GW by 2030 against roughly 17 GW of demand by 2031-32.
- About USD 70 billion has been invested with a further USD 90 billion announced; data centres have held infrastructure status since the 2022-23 Budget.
- The 2026-27 Budget adds a tax holiday for eligible foreign cloud service providers through 2047, aimed at pulling hyperscale capacity onshore.
- The binding constraints are now power, land and water - plus the imported silicon inside - not demand; the gap between announced and built capacity is the execution test.
- Press Information Bureau / MeitY (14 Sep 2026): https://www.pib.gov.in/PressReleasePage.aspx?PRID=2309950
- JLL India Data Centres 2026 mid-year report: https://www.jll.com/en-in/insights/market-dynamics/india-data-centers
- Communications Today (2026): https://www.communicationstoday.co.in/india-data-centre-2026-mid-year-report/
The development
A MeitY feature published on 14 September 2026 put hard numbers on India's data-centre build-out. Installed capacity has risen from around 375 MW in 2020 to 1.57 GW as of August 2026, with the government projecting nearly 8 GW by 2030 against demand of roughly 17 GW by 2031-32. Cumulative investment stands at about USD 70 billion, with a further USD 90 billion in projects announced. The state operates four National Data Centres - at Delhi, Pune, Hyderabad and Bhubaneswar - alongside 37 smaller facilities in state capitals. Data centres have held infrastructure status since the 2022-23 Union Budget, and the 2026-27 Budget adds a tax holiday for eligible foreign cloud service providers running through 2047.
Why it matters
The policy centre of gravity has shifted from whether India can host compute to how fast it can pull hyperscale capacity onshore. Keeping AI and cloud workloads inside the country is a data-sovereignty, latency and cost question at once, and the tax holiday for foreign cloud providers is a deliberate lever to convert announced intent into built capacity. The government is also explicitly wiring data centres to its other industrial bets - the Rs 1,27,500 crore ISM 2.0, the IndiaAI Mission, an enlarged ECMS and PLI 2.0 for IT hardware - treating compute as strategic infrastructure rather than a commercial real-estate line.
The dependency and systems frame
Capacity measured in gigawatts is the visible number; the constraints sit underneath it. Eight gigawatts of data-centre load is first a power problem - reliable supply, grid connections and the clean-energy share - then a land and water problem, since cooling at that scale is water-intensive. And the silicon inside the halls - accelerators, networking and memory - remains overwhelmingly imported, so a larger domestic compute footprint does not by itself reduce the deepest import dependency. The gap between USD 90 billion announced and 1.57 GW actually built is the honest measure of where the sector is.
What to watch
- Whether the announced USD 90 billion pipeline and the ~8 GW-by-2030 projection convert into built, powered capacity on schedule.
- Uptake of the 2026-27 budget tax holiday by foreign cloud and hyperscaler investment.
- Power sourcing for data-centre load - grid reliability and how much is met from clean energy.
- Whether domestic content in the stack (chips, networking, cooling) deepens, or capacity growth stays import-reliant.
Track the systems we watch
Signals, reports and briefings on India’s industrial transformation.