Hyperscale data centres — the backbone of global cloud computing and artificial intelligence — are no longer a benign infrastructure choice for developing nations. These are not modest server rooms. A single campus can consume as much electricity as a mid-sized power plant, draw millions of litres of water daily, and occupy large tracts of land, all while employing fewer than a hundred people. The economic value they generate flows almost entirely to multinational technology companies headquartered far from the countries bearing the resource costs. Several governments across the world have already responded with bans, moratoria, or strict regulatory controls. For India — a country ranked 134th on the UN Human Development Index with acute per capita electricity and water deficits — the case for regulatory action is stronger than for any of them.

Hyperscale data centres are the heavy industry of the digital era — but unlike traditional industry, they generate minimal employment while consuming vast quantities of electricity and water.

 The Resource Burden: What These Facilities Actually Consume

Electricity: Unrelenting Baseload Demand

A single hyperscale campus requires 100–500 MW of continuous, uninterruptible power. Large AI computing clusters can demand 1 GW or more — equivalent to a dedicated mid-sized power plant running at full capacity, reserved for a single private facility. Unlike most industrial loads, this demand cannot be curtailed during peak periods; it must be met regardless of grid conditions, weather, or competing demand. Global data centre electricity consumption is projected to more than double by 2030, driven predominantly by AI workloads.

For grids already under stress, the consequences are concrete: industrial and manufacturing demand gets crowded out as capacity is absorbed; system-wide infrastructure costs rise and are spread across all consumers; electricity tariffs increase for households and businesses that had nothing to do with the investment decision; and fossil fuel backup generation must be maintained to serve the constant load profile, even when operators nominally hold renewable energy certificates. Electricity consumed serving foreign workloads is, in effect, an invisible energy export — one that returns far less economic value than conventional exports while leaving the host nation bearing every infrastructure cost.

Water: A Cost Hidden in Plain Sight

Cooling high-density computing equipment requires between 10 and 20 million litres of water per day at a single large facility — billions of litres annually, comparable to a mid-sized city. In water-stressed regions this is not an abstract environmental metric; it is direct competition with municipal supply and agricultural irrigation. Unlike manufacturing or agro-processing, this water consumption generates negligible employment or domestic economic multipliers.

Employment and Economic Value: The Structural Gap

Hyperscale facilities are capital-intensive by design and lean on labour by necessity. Automation and operational efficiency are central to their profitability. A single operational campus typically employs between 30 and 100 permanent staff — a remarkably small workforce for an asset consuming hundreds of megawatts of electricity and billions of litres of water annually. Compare this with a manufacturing facility or logistics hub consuming equivalent power, which might employ thousands.

When these facilities primarily serve foreign workloads — global cloud platforms, AI computing for non-domestic users — the economic case for the host country weakens further. Operating profits accrue almost entirely to multinational technology companies. Domestic supply-chain spillovers, skills transfer, and local vendor development remain limited. Digital service exports generated through these facilities do not build local technology capability. The host country supplies the electricity, water, land, and public grid infrastructure. The value is captured elsewhere.

 International Precedents: Who Has Already Acted and Why

The following jurisdictions have each introduced restrictions after recognising that unrestricted hyperscale growth imposes unacceptable strains on national infrastructure. What is particularly striking is that all of them have substantially greater per capita electricity generation, grid maturity, and water availability than India — and they still concluded that the trade-off was not worth making.

The HDI Dimension: A Question of Resource Equity

The debate over hyperscale data centres is typically framed in macro terms — aggregate national grid capacity, total water availability, projected GDP contribution. These are valid measures, but they obscure a more fundamental question: how does this resource allocation affect the welfare and development prospects of ordinary citizens? The Human Development Index measures outcomes that matter directly to people — life expectancy, education, and income. When examined through this lens, India's position is not just distinctive; it makes the case for regulatory action unanswerable.

Per Capita Electricity: A Development Deficit

India's per capita electricity consumption stands at approximately 1,300 kWh per year — among the lowest in the G20 and less than half the global average of roughly 3,500 kWh. High-HDI countries average above 6,000 kWh per capita annually; the United States exceeds 12,000 kWh. This is not merely a statistical gap. It represents unmet need that is visible in healthcare facilities running on intermittent power, schools without consistent digital access, cold chains that cannot function, and rural households locked out of productive economic activity. Research consistently demonstrates that below approximately 4,000 kWh per capita, increases in electricity access produce measurable improvements in child mortality rates, literacy, life expectancy, and household income — outcomes that define the HDI itself.

A 300 MW hyperscale campus consumes approximately 2.6 billion kWh annually — equivalent to the electricity entitlement of over 2 million Indians at current consumption levels. A 1 GW AI cluster equals the residential consumption of roughly 7 million people. These are not abstract numbers. They represent a concrete development trade-off that India's policy framework currently treats as invisible.

India is ranked 134th on the UNDP HDI. The countries imposing data centre restrictions — Singapore, Ireland, the Netherlands — are ranked in the top 25. If they concluded that national infrastructure must serve national development, the logic applies with far greater force to India.

Per Capita Water: Scarcity in Real Terms

India's per capita freshwater availability has fallen from over 5,000 cubic metres at independence to approximately 1,400 cubic metres today — approaching the internationally recognised water-stress threshold of 1,700 m³. Maharashtra, Rajasthan, Karnataka, and Tamil Nadu are already classified as water-scarce. The WHO recommends a minimum of 50 litres per person per day for basic health and hygiene; hundreds of millions of Indians receive less during summer months.

A single large hyperscale facility consuming 15 million litres per day uses water equivalent to the WHO daily minimum for 300,000 people — every day. Water access is among the most direct determinants of HDI outcomes: inadequate supply drives disease burden, reduces agricultural productivity, and keeps women and girls out of education. Allocating this water to servers processing foreign digital workloads is not a neutral resource decision. It is a choice to prioritise the convenience of high-income foreign users over the basic welfare of Indian citizens.

 A Rational Policy Framework

The objective is not to restrict digital infrastructure categorically — domestic digital capability is essential to India's development. The objective is to distinguish infrastructure that serves national priorities from infrastructure that extracts value from national resources while delivering it elsewhere. A workable framework rests on a clear two-category distinction:

Domestic Digital Infrastructure — Support and Prioritise

Facilities serving government platforms, financial systems, domestic cloud demand, and India's digital public infrastructure should receive streamlined approvals and active policy support. They build domestic capability and generate value that circulates within the economy.

Export-Oriented Hyperscale — Regulate Strictly or Prohibit

Facilities primarily serving foreign workloads should face mandatory conditions: captive renewable generation (not paper certificates), full cost recovery for dedicated grid infrastructure, prohibition on potable water use for cooling, and strict energy efficiency standards. In water-stressed or power-deficit regions, a formal moratorium is the appropriate starting point — consistent with what Singapore, Ireland, and the Netherlands have already done from a far more comfortable resource position.

 Conclusion

The countries that have restricted hyperscale data centres — Singapore (ranked 9th on the HDI), Ireland (8th), the Netherlands (10th) — are wealthy, resource-comfortable, and operate at the very top of global human development rankings. They imposed restrictions because they concluded that national infrastructure must serve national development priorities, not the operational preferences of global technology companies. If that logic was compelling for nations with ten times India's per capita electricity consumption and comfortable water availability, it applies with overwhelming force to a country ranked 134th on the HDI, where per capita electricity consumption stands at roughly 1,200 kWh and freshwater availability is approaching internationally recognised stress thresholds.

Permitting unrestricted hyperscale expansion serving non-domestic demand is not a neutral infrastructure policy. It is a decision to export India's electricity and water through digital services while capturing minimal economic value — effectively subsidising global technology platforms at the cost of domestic development. A country with India's resource constraints and development obligations cannot afford that trade-off.