Economic Survey flags power, water, GPU bottlenecks as India targets 4GW data centre capacity

India’s data-centre capacity is projected to more than triple to about 4 gigawatts (GW) by 2030, driven by cloud adoption, artificial intelligence (AI), IoT and 5G, but the Economic Survey 2025–26 warned that energy, water and hardware constraints could slow the expansion.

Installed capacity stood at around 1,280 megawatts (MW) as of June 2025, with about 130 privately operated data centres and 49 run by central and state government agencies. The sector remains largely private-led and deregulated, supported by Make in India and Atmanirbhar Bharat initiatives to strengthen electronics and semiconductor manufacturing across the value chain, the Survey said.

To support public digital services, the government has also expanded MeghRaj, its cloud platform, empanelling 26 cloud service providers as of December 2025 to accelerate e-governance, digital payments and consent-based data sharing.

However, the Survey highlighted mounting physical pressures from AI-led infrastructure growth. Data centres require large volumes of electricity and water, while AI workloads introduce volatility into power demand, risking grid stability.

India currently hosts about 3% of global data centres by count, compared with 73% in high-income countries, despite generating nearly 20% of the world's data.

"With emerging hubs such as Malaysia (Johor), Japan, and Vietnam intensifying competition, addressing structural constraints such as energy shortages will be critical for India to position itself as a global AI data centre hub," the Survey said.

The Economic Survey said India must be careful about expanding AI infrastructure, as power, finance and especially water remain limited. It warned that building large-scale compute capacity comes with trade-offs, since resources used for AI also compete with the needs of households and industry. 

“For India, as power, finance, and especially water remain binding constraints, scaling compute indiscriminately carries opportunity costs. Investment in AI infrastructure competes directly with other sources of demand, such as households and industries. This creates a trade-off between centralised scale and distributed efficiency, strengthening the case for smaller, task-specific models that can run on limited hardware and decentralised compute networks,” the Survey said.

Hardware availability is emerging as the sharpest constraint. Surging demand for GPUs and shortages of high-bandwidth memory chips are already driving cost pressures, with the Survey warning that even well-funded projects may be delayed until supplies are secured.

"The absence of a steady supply of semiconductors is likely to hamper economic activities across sectors," the Economic Survey said.