Data centre capacity in the country has increased from about 375 MW in 2020 to around 1500 MW by 2025

1. At a Glance

• India's installed data centre (DC) capacity quadrupled from ~375 MW (2020) to ~1500 MW (2025), anchoring the country's digital infrastructure stack for cloud, 5G, and AI workloads [1].
• DC growth is now tightly coupled with the IndiaAI Mission's compute layer — 38,231 GPUs onboarded via 14 empanelled service providers at a subsidised ₹65/hour (~1/3 of global average) [1][3].
• UPSC relevance: cross-cuts GS-III (Infrastructure, IT, AI), digital sovereignty, data protection (DPDP Act 2023) and energy/water sustainability debates.

2. Why in the News

• PIB release dated 13 March 2026 by Ministry of Electronics & IT (MeitY) quantified the 375 MW → 1500 MW jump and detailed adoption of advanced cooling technologies to cut water use [1].
• IndiaAI Mission crossed 34,000+ GPUs (June 2025) and then 38,000+ GPUs in the AI Compute Portal, with the second empanelment round expanding capacity [2][3].

3. Background & Evolution

• 2020: ~375 MW installed capacity; sector recognised as infrastructure by MoF (March 2022 Budget grant of infra status to DC) [1].
• 2020-22: TRAI Recommendations (Nov 2022) on a regulatory framework for Data Centres, CDNs and Internet Exchanges, proposing a Data Centre Incentivization Scheme (DCIS) and Data Centre Economic Zones (DCEZs) [4].
• 2023: Draft National Data Centre Policy targets +2000 MW addition by 2027; BIS tasked with India-specific DC building standards [5].
• Aug 2024: MeitY (via IndiaAI IBD) issued Request for Empanelment (RFE) for AI compute/cloud providers on CPP portal [3].
• 2025: 14 providers empanelled, 38,231 GPUs operational; DC capacity touches ~1500 MW [1][3].

4. Core Static Facts

• Nodal ministry: Ministry of Electronics & Information Technology (MeitY) [1].
• Implementing arm for AI compute: IndiaAI Independent Business Division (IBD) under Digital India Corporation [3].
• Capacity: 375 MW (2020) → ~1500 MW (2025); target +2000 MW by 2027 (draft National DC Policy) [1][5].
• GPUs onboarded: 38,231 via 14 empanelled providers; subsidised rate ₹65/hr [1].
• Beneficiaries: startups, MSMEs, academia, researchers, PhD scholars, government agencies [3].
• Hubs: Mumbai, Navi Mumbai, Hyderabad, Bengaluru, Noida, Jamnagar [1].
• Enabling instruments: Draft National Data Centre Policy; TRAI 2022 recommendations; DPDP Act, 2023 (data localisation backdrop); SEZ → DCEZ conversion proposals for AP, Kerala, Karnataka, Maharashtra, Haryana, UP, MP, Gujarat, Rajasthan, Odisha [4][5].
• Standards body: Bureau of Indian Standards (BIS) — DC construction & certification framework [5].

5. Multi-Dimensional Analysis

Economic

• Sector classified as infrastructure → access to long-tenure infra debt and harmonised master list [5].
• Subsidised compute (₹65/hr ≈ 1/3 global) lowers entry barriers for Indian AI startups [1].

Scientific / Technological

• Shift to liquid cooling, direct-to-chip cooling and free-air cooling to slash water/PUE; relevant for water-stressed hubs like Hyderabad/Chennai [1].
• GPU pool supports foundational model training under IndiaAI Mission [2][3].

Environmental

• DCs are power- and water-intensive; 1 MW DC ≈ 25-30 million litres/yr water in traditional cooling — MeitY flags shift to advanced cooling [1].
• Tension with India's net-zero by 2070 commitment; push for renewable PPAs and green DCs [5].

Geopolitical / Strategic

• Digital sovereignty: domestic compute reduces reliance on US/EU hyperscalers for sensitive workloads [2].
• Counters China's BRI digital silk-road via indigenous AI stack [3].

Administrative / Federal

• State Data Centre (SDC) scheme (MeitY) since 2008 funds Tier-III SDCs in every State/UT for e-Gov [6].
• DCEZ proposal requires Centre-State coordination on SEZ Act amendments, land, power tariffs [4].

6. Recent Developments (last 12-18 months)

• Aug 2024: RFE issued by IndiaAI IBD for GPU empanelment [3].
• 2024-25 (Round 1+2): GPU pool grew 18,417 → 34,333 → 38,231 [2][1].
• 2025: IndiaAI Mission rolled out compute access for startups at ₹65/hr [3].
• 13 March 2026: PIB confirms 1500 MW milestone and water-efficient cooling adoption [1].

7. Prelims Hooks

• DC capacity rose from ~375 MW (2020) to ~1500 MW (2025) [1].
• 38,231 GPUs onboarded under IndiaAI compute framework [1].
• Number of empanelled AI service providers/DCs: 14 [1].
• Subsidised GPU rate: ₹65/hour, ≈ one-third of global average [1].
• Six DC hubs named in PIB: Mumbai, Navi Mumbai, Hyderabad, Bengaluru, Noida, Jamnagar [1].
• Nodal ministry: MeitY (not MoCommunications, not MoF) [1].
• Implementing body for AI compute: IndiaAI IBD under Digital India Corporation [3].
• TRAI issued data-centre regulatory framework recommendations in November 2022 [4].
• Draft National Data Centre Policy targets +2000 MW by 2027 [5].
• DC sector granted infrastructure status for harmonised infra lending [5].
• BIS is developing India-specific DC building standards [5].
• DCEZ = Data Centre Economic Zones (SEZ conversion proposal) [4].
• State Data Centre (SDC) scheme is run by MeitY [6].

8. Mains Relevance

• GS-III — Infrastructure: Energy, Ports, Roads, IT; Science & Tech – Indigenisation; Awareness in IT, AI.
• GS-II — Government policies & interventions; e-Governance.
• Probable stems: 1. "Data centres are the 'factories of the AI economy'. Examine India's policy framework to make the country a global DC hub, highlighting environmental trade-offs." (GS-III) 2. "Discuss how the IndiaAI Mission's compute capacity layer addresses the digital sovereignty challenge." (GS-III) 3. "Evaluate the rationale for granting infrastructure status to data centres and converting SEZs into DCEZs." (GS-III)

9. Related Topics to Study Next

• IndiaAI Mission (2024) — parent programme for the GPU empanelment [3].
• DPDP Act, 2023 — data localisation drives DC demand.
• Semiconductor Mission (ISM) & DLI scheme — hardware backbone for AI compute.
• National Broadband Mission / Bharat 6G Vision — connectivity layer feeding DCs.
• TRAI recommendations on CDNs & IXPs (2022) — companion regulatory piece [4].
• State Data Centre scheme — federal e-Gov backbone [6].
• Climate commitments (Panchamrit, LiFE) — DC energy/water footprint debate.
• GCC (Global Capability Centres) growth in India — co-located DC demand driver.

10. Common Errors / Trap Areas

• Confusing MeitY (DC/AI compute) with DoT (telecom/CDNs) — TRAI advises both, but MeitY is nodal for DCs [1][4].
• Mixing the State Data Centre (SDC) scheme with commercial hyperscale DCs — SDC is for State e-Gov; PIB's 1500 MW is total industry capacity [1][6].
• The ₹65/hr is the subsidised compute rate, not an electricity/colocation tariff [1].
• DCEZ ≠ SEZ — DCEZ is a proposed class under draft policy/TRAI, not yet legislated [4].
• DC infrastructure status was conferred via Union Budget (Harmonised Master List), not by a standalone Act [5].