Challenges for India’s Electricity Grid from Data Centre Expansion

India’s Data Centres and the Looming Electricity Grid Crisis

By 2030, India’s installed data centre capacity is projected to expand nearly tenfold, from 1.2 GW today to 10 GW, attracting investments exceeding $200 billion. That’s the opportunity. But the challenge? The growing computational intensity of artificial intelligence (AI) workloads that demand energy at levels our electricity grid seems poorly prepared to deliver. Individual AI racks require 80-150 kW of electricity—several times that of traditional enterprise servers. While these data centres operate 24/7 with a stable base load, the sudden spikes during peak AI processing could strain grid flexibility to its breaking point.

Why the Data Centre Boom Breaks the Energy Pattern

India’s energy sector has traditionally revolved around predictable consumption patterns—from industrial zones, metro rails, irrigation pumps, and urban households. But AI-driven data centres are an entirely different beast. Unlike predictable loads, their demand surges are both colossal and erratic. This fundamentally challenges grid management. The timeline for this transition compounds the difficulty: the leap from 1.2 GW to 10 GW in just four years (2026 to 2030) is not gradual evolution; it’s a grid-shattering disruption.

Countries like the US, facing similar growth trajectories, have spent decades preparing through targeted investments in “energy-resilient data hubs.” India does not have the luxury of time. Worse, systemic complacency in upgrading sub-transmission infrastructure risks turning these hyperscale facilities into flashpoints of local grid failures. Despite the government’s ambitious renewable energy targets, there remains minimal discussion of aligning these with the energy-intensive trajectory of digital infrastructure.

The Institutional Machinery at Play

The Central Electricity Authority (CEA), responsible for electricity planning under the Electricity Act, 2003, has consistently underestimated the sectoral composition of demand. Current national energy plans do not fully anticipate the dual challenge of providing consistent baseload power while accommodating frequent, high-demand spikes unique to AI computing workloads. Moreover, Transmission Corporations (Transcos) are unequipped to develop the ultra-high-voltage substations and additional high-capacity corridors necessary to deliver uninterrupted power to hyperscale facilities.

Even at the planning level, data centres remain an afterthought. The broader problem, as observed in India’s shift to renewables too, is the absence of updated grid codes to address dynamic, unpredictable loads. Current guidelines issued by state regulators focus overwhelmingly on traditional industrial and urban demand patterns rather than incorporating AI-specific energy demands. Promising buzzwords like “smart grid systems” have made their way into policy documents, but implementation remains nebulous, and coordination between the Power Ministry and state-level electricity boards continues to lag.

The Data Problem Beneath the Power Problem

The government claims India is ready to meet digital energy demands through increased baseload capacity and renewables integration. But specific data tells another story. Despite boasting a renewable installed capacity of 127 GW, of which solar energy constitutes roughly 71 GW, only about 25% of this is dispatchable — that is, capable of meeting peak demand variations reliably. For AI-driven centres, the ability to draw stable power—even during sudden consumption surges—will be critical. There’s little in the planning documents to show that this gap is being acknowledged, let alone addressed.

The reality is also geographic. AI data centre projects are concentrated in urban hubs—Mumbai, Bengaluru, Chennai, Hyderabad—where existing sub-transmission systems are already grappling with urban sprawl. The mismatch is stark: between 2015 and 2023, urban electricity demand in these regions grew at an annual pace of 6-8%, whereas regional transmission capacity additions have barely crossed 3% annually.

The difficulty also lies in resource adequacy during demand surges. States like Tamil Nadu, heavily reliant on wind energy, may find themselves struggling during windless nights. Coupling these challenges with the budgetary constraints of state electricity boards (SEBs)—many of which are already in financial distress—paints a bleak picture of preparedness.

The Governance Questions Nobody Asks

The government’s framing of this issue as a power capacity challenge is, at best, superficial. The real questions are structural. What are the safeguards against regulatory capture when data centre developers push for subsidised energy tariffs? Should SEBs—which already receive bailout packages annually—bear the cost of upgrading the grid to cater to hyperscale developers? And where will the funding come from for ultra-high-voltage transmission corridors essential for transporting bulk power to concentrated data hubs?

Consider also the policy asymmetry in promoting renewables. While the Centre has set aggressive targets for installing 500 GW of non-fossil fuel capacity by 2030, the political economy of coal still dominates baseload generation. Can India realistically phase in low-carbon, reliable baseload sources—hydro and nuclear—fast enough to meet AI-driven electricity demand?

Finally, federal slipstreams matter. Energy policy, particularly grid planning, is largely a concurrent subject. Yet, states operate energy markets in practice. The lack of coordination between central agencies like the CERC (Central Electricity Regulatory Commission) and state electricity regulatory commissions is already evident in prolonged disputes over tariff agreements and project clearances. AI-driven data centre energy demands will only accentuate these existing fault lines.

Lessons From South Korea

When South Korea faced a similar boom in hyperscale data centres in 2018, it responded with targeted, scalable action. The government mapped out “Energy-IT Corridors” across underdeveloped regions like Gyeongnam, integrating renewable energy hubs directly with cluster-specific grid upgrades. Crucially, it established direct policy mechanisms linking energy subsidies for hyperscale facilities to contributions towards local grid infrastructure upgrades. In contrast, India lacks region-specific grid integration plans for its data centres, with projects clustering unplanned in overburdened urban areas prone to grid bottlenecks.

Practice Questions for UPSC

Prelims Practice Questions

Frequently Asked Questions