New Delhi.
India’s digital transformation journey is accelerating at an unprecedented scale, propelled by the nationwide expansion of artificial intelligence, high-density graphics processing unit (GPU) computing clusters, and ambitious sovereign AI initiatives. However, behind the massive capital investments and headline-grabbing mega-projects lies a complex operational reality. As tech hyperscalers, telecom operators, and real estate developers scale up computing capacity, India’s data-centre industry is hitting a critical juncture: the pressing challenge of power supply, network connectivity, and physical grid infrastructure.
The Power Crunch: From Silicon to Energy
The global AI bottleneck has fundamentally shifted from chip shortages to power distribution. Standard cloud workloads operate on predictable power draws, but high-density AI clusters running deep learning models demand non-stop, 24/7 high-load electricity.
┌──────────────────────────────────────────────────────────┐
│ INDIA’S DATA CENTRE POWER EVOLUTION │
├──────────────────────────────────────────────────────────┤
│ Current Capacity (2026) : ~2.4 GW │
│ Projected Target (2030) : 9.0 GW – 12.0 GW │
│ Primary Grid Challenge : Transmission & Delivery (HVDC)│
└──────────────────────────────────────────────────────────┘
- Grid Delivery Limitations: In top data-centre clusters like Mumbai—which holds roughly 30% of national capacity—the core obstacle is not power generation, but urban transmission and delivery. High-voltage direct current (HVDC) underground transmission lines are being deployed to overcome urban dense-space bottlenecks.
- Clean Energy Integration: Hyperscalers are aiming for Net-Zero operations, necessitating massive Power Purchase Agreements (PPAs) for solar and wind energy. To combat renewable intermittency, developers are turning to large-scale Battery Energy Storage Systems (BESS).
Network Connectivity & Subsea Interconnects
AI models rely on multi-terabit datasets distributed across interconnected nodes. High bandwidth and ultra-low latency are non-negotiable requirements for AI inference and continuous cloud synchronization.
- Subsea Cable Diversification: To avoid bottlenecking on traditional West Coast landing stations in Mumbai, major operators are establishing East Coast landing gateways in Visakhapatnam (Vizag) and Chennai.
- Dark Fiber Corridors: State-to-state latency relies heavily on direct, low-loss dark fiber backbones connecting secondary computing hubs directly to high-capacity subsea gateways.
Real Estate, Water Constraints, and Regionalization
Geographic concentration in Tier-1 cities has inflated land costs (reaching upwards of ₹30+ crore per acre in prime metro locations) and strained municipal resources.
- Liquid Cooling Adoption: Air cooling is no longer efficient for racks pulling over 40–50 kW per unit. Developers are rapidly upgrading to direct-to-chip liquid cooling and closed-loop immersion setups to manage thermal profiles while conserving local municipal water supplies.
- Tier-2 Expansion: To circumvent land and grid constraints, developers are expanding into secondary corridors across Maharashtra, Gujarat, Telangana, Andhra Pradesh, and Uttar Pradesh.
Frequently Asked Questions (FAQ)
Q1: Why is power supply becoming the primary bottleneck for AI data centres in India?
AI training models require continuous, high-density load power without fluctuations. Delivering thousands of megawatts to localized data-centre parks strains urban electrical distribution networks, requiring major upgrades like High-Voltage Direct Current (HVDC) power lines.
Q2: How are data centre developers addressing the water consumption challenge?
Traditional evaporative cooling uses substantial water volumes. Modern facilities are transitioning to direct-to-chip liquid cooling and closed-loop immersion cooling, drastically reducing water usage while maintaining GPU efficiency.
Q3: Which cities are emerging as new data-centre hubs in India?
While Mumbai and Chennai remain primary markets, rapid expansion is occurring in secondary and coastal hubs like Visakhapatnam, Hyderabad, Pune, Noida, and parts of Gujarat.
Disclaimer
The information, analysis, and views presented in this article are compiled for informational and educational purposes only. Market projections, data-centre capacity figures, and policy details reflect industry estimates and public developments current as of 2026. Readers and prospective investors are advised to perform independent research and verify official regulatory or market filings before making strategic business decisions.
Readings
Visit Bharat Kaalvrutt.
