India’s AI Data-Centre Opportunity Faces Power-Supply Challenge as Sovereign AI and Cloud Demand Accelerate

India could become one of the biggest beneficiaries of the global artificial-intelligence data-centre investment boom, as demand for sovereign AI, cloud computing, data localisation and GPU-based infrastructure pushes companies to deploy significantly more computing capacity inside the country.

But the scale of the opportunity is creating an equally significant infrastructure challenge:

electricity.

AI data centres consume substantially more power than conventional computing facilities, particularly when operating thousands of high-performance graphics processing units used for training and running large AI models.

India's data-centre capacity has already expanded from approximately 375 MW in 2020 to around 1,575 MW in 2026, according to government data.

The next phase could be dramatically larger.

The Central Electricity Authority estimates that data-centre power requirements could reach around 17 GW by 2031-32, while another government projection has estimated an additional 26.3 GW of load from AI data centres alone by that period.

Industry forecasts are even more aggressive on physical capacity. Wood Mackenzie expects India's operational data-centre capacity to increase from approximately 2.2 GW in 2025 to 12 GW by 2030.

The figures point toward a fundamental change in India's digital economy.

Data centres are moving from being primarily technology infrastructure to becoming a major component of national energy and industrial planning.

India Is Emerging as a Major AI Data-Centre Market

Several structural factors are pushing data-centre investment toward India.

The country combines:

a population exceeding 1.4 billion,

rapid digital adoption,

one of the world's largest internet-user bases,

expanding cloud consumption,

growing enterprise AI adoption,

government-backed AI infrastructure,

and increasing data-localisation requirements.

Generative AI adds another powerful demand driver.

Companies that previously relied primarily on conventional cloud servers increasingly require specialised GPU infrastructure capable of running computationally intensive AI workloads.

That shift dramatically increases both the amount and density of computing equipment deployed inside data centres.

Sovereign AI Is Creating New Infrastructure Demand

One of the strongest emerging themes is sovereign AI.

Governments and enterprises increasingly want strategically important AI workloads, models and data to remain within domestic infrastructure.

Sovereign AI can involve domestic control over several layers of the technology stack:

energy,

semiconductors,

data centres,

foundation models,

applications,

data,

and talent.

India's AI strategy increasingly reflects this approach.

The objective is not simply to use artificial intelligence developed elsewhere.

India wants greater domestic capability to train, host and deploy AI systems using infrastructure located within the country.

That creates significant demand for local computing capacity.

IndiaAI Mission Is Expanding Domestic GPU Access

The IndiaAI Mission is one of the government's major initiatives supporting this infrastructure expansion.

India has already onboarded tens of thousands of GPUs through empanelled compute providers.

By March 2026, approximately:

38,231 GPUs

had been onboarded through 14 service providers and data centres under the AI compute-capacity framework.

The infrastructure is being made available to:

startups,

researchers,

academic institutions,

government organisations,

and other eligible users.

Subsidised access is intended to reduce one of the largest barriers to AI development: the high cost of computing.

AI Compute Is Becoming Strategic Infrastructure

AI models require enormous computational resources.

Training a sophisticated model may involve thousands of GPUs operating continuously for weeks or months.

Once a model is deployed, inference workloads create continuing computing demand every time users interact with it.

As AI applications expand across:

banking,

healthcare,

manufacturing,

education,

commerce,

government,

media,

and enterprise software,

the underlying infrastructure must scale with them.

This is turning GPU computing into an infrastructure category comparable in strategic importance to telecommunications and cloud computing.

Global Cloud Companies Are Increasing India Investments

International technology companies are committing large amounts of capital to Indian digital infrastructure.

Major announced investment programmes include substantial commitments from global cloud and technology companies for:

data centres,

cloud infrastructure,

AI computing,

digitalisation,

and workforce development.

India's appeal is increasingly based on more than domestic demand.

The country could also become a location from which AI computing services are supplied to international customers.

That would transform Indian data centres from primarily domestic infrastructure into export-oriented digital infrastructure.

Yotta Highlights India's Sovereign Cloud Opportunity

Indian data-centre companies are also positioning themselves for the AI expansion.

Yotta Data Services has developed large-scale Nvidia-powered AI infrastructure and is expanding sovereign cloud capabilities.

The company has indicated that a substantial majority of its customers are international.

That illustrates a potentially important shift.

India may not only consume AI infrastructure.

It could export computing capacity.

If power, connectivity and cost structures remain competitive, international companies facing infrastructure constraints elsewhere could deploy workloads in Indian data centres.

Power Could Become India's Biggest Constraint

The central challenge is that AI infrastructure requires enormous quantities of reliable electricity.

A conventional data centre already operates continuously.

AI facilities can require even greater power density because GPU servers consume considerably more electricity than traditional computing hardware.

An AI data centre cannot tolerate frequent interruptions.

It requires:

continuous electricity,

stable voltage,

backup generation,

redundant power systems,

and increasingly large grid connections.

This means the issue is not simply whether India generates enough electricity nationally.

Power must be available reliably at the specific locations where large data centres are constructed.

Government Estimates Data-Centre Power Demand at 17 GW

Government data illustrate the potential scale of the challenge.

India currently has approximately:

1,575 MW of installed data-centre capacity.

The Central Electricity Authority has estimated that data-centre power requirements could rise to around:

17 GW by 2031-32.

That would represent a dramatic increase from current levels.

Another government estimate has projected that AI data centres could add approximately 26.3 GW of load by FY32.

The different estimates reflect varying definitions and rapidly changing expectations around the pace of AI infrastructure development.

But the direction is consistent:

data-centre electricity demand is set to rise sharply.

Wood Mackenzie Forecasts 12 GW of Capacity by 2030

Industry projections also point toward extraordinary expansion.

Wood Mackenzie expects India's operational data-centre capacity to rise from approximately:

2.2 GW in 2025

to:

12 GW by 2030.

That would represent annual growth of roughly:

40%.

AI-dedicated capacity is expected to grow even faster.

Wood Mackenzie projects AI-specific capacity could increase almost 24-fold by 2030.

This changes the economics of the entire sector.

For future data-centre developers, access to electricity could become as important as access to land and fibre connectivity.

Electricity Strategy Could Become a Competitive Advantage

Historically, data-centre developers evaluated locations using factors such as:

land availability,

fibre connectivity,

proximity to customers,

tax incentives,

and regulatory policy.

Electricity is increasingly moving to the top of that list.

A location with abundant land but insufficient grid capacity may no longer be commercially attractive.

Developers need certainty that hundreds of megawatts of additional electricity can be supplied on schedule.

Large AI campuses may eventually require power comparable with major industrial facilities.

States capable of providing reliable, affordable and increasingly low-carbon electricity could therefore capture a disproportionate share of future investment.

Mumbai Remains India's Largest Data-Centre Hub

Mumbai and Navi Mumbai remain the country's largest data-centre cluster.

The region benefits from:

international submarine cables,

financial-services demand,

large enterprise customers,

cloud infrastructure,

and established data-centre ecosystems.

Other major hubs include:

Chennai,

Hyderabad,

Bengaluru,

Delhi NCR,

Pune,

and Jamnagar.

However, continued expansion in established markets could become increasingly difficult if power, land and water constraints intensify.

This is encouraging developers to examine new regions.

New States Are Emerging as Data-Centre Destinations

India's data-centre industry is beginning to expand beyond its traditional metropolitan clusters.

Government data indicate growing investment interest in states including:

Andhra Pradesh,

Madhya Pradesh,

Chhattisgarh,

and West Bengal.

New locations can offer:

larger land parcels,

lower real-estate costs,

greater access to power,

renewable-energy potential,

and supportive state incentives.

The rise of AI could accelerate this geographic diversification.

Unlike latency-sensitive consumer applications, some AI training workloads can be located farther from major population centres if sufficient power and connectivity are available.

AI Training Could Change Data-Centre Geography

AI training and conventional cloud computing do not always have identical infrastructure requirements.

Certain cloud applications need to operate extremely close to users to minimise latency.

Large AI training workloads can sometimes tolerate greater physical distance.

This creates an opportunity to build large AI campuses closer to:

renewable-energy projects,

high-capacity transmission infrastructure,

industrial zones,

and lower-cost land.

India could therefore develop a new generation of data-centre hubs based primarily on energy availability rather than proximity to major cities.

Renewable Energy Will Be Essential

Meeting rapidly rising data-centre electricity demand entirely through fossil fuels would create significant environmental and policy challenges.

Global technology companies have ambitious carbon-reduction targets.

Many therefore want their data centres to operate using low-carbon electricity.

India's renewable-energy expansion could become a major competitive advantage.

The country has substantial potential across:

solar,

wind,

hydropower,

energy storage,

and eventually expanded nuclear generation.

But renewable generation alone does not solve the data-centre power problem.

AI Data Centres Need Electricity 24 Hours a Day

Solar power is available primarily during daylight hours.

Wind generation varies according to weather conditions.

AI servers, however, can operate continuously.

A hyperscale data centre may require large quantities of electricity:

24 hours a day, 365 days a year.

This creates a mismatch between variable renewable generation and continuous computing demand.

The solution requires a combination of:

renewable generation,

battery storage,

pumped hydro,

grid balancing,

transmission infrastructure,

and reliable firm power.

Energy planning therefore becomes inseparable from AI infrastructure planning.

Grid Infrastructure Is as Important as Generation

India can have sufficient electricity generation nationally while still experiencing local constraints.

Power must be transported from generation sites to data-centre clusters.

That requires:

high-voltage transmission lines,

substations,

transformers,

distribution infrastructure,

and grid interconnections.

These projects often require years of planning and construction.

A data-centre campus can sometimes be developed faster than the supporting transmission infrastructure.

If grid expansion does not occur simultaneously, electricity availability could delay new facilities.

India's Wider Grid Already Faces Growing Demand

The data-centre challenge is occurring while India's overall electricity demand is also expanding.

Industrialisation,

air conditioning,

electric vehicles,

urbanisation,

railway electrification,

manufacturing,

and rising household consumption

are all adding load to the grid.

Data centres therefore compete for electricity within a much larger national growth story.

This does not mean India lacks generation capacity.

It means electricity planning must account for several rapidly expanding sources of demand simultaneously.

Storage Will Become Increasingly Important

Energy storage can help bridge the gap between renewable generation and continuous data-centre demand.

India is planning major expansion of:

battery energy-storage systems

and

pumped-storage projects.

Government plans envisage approximately:

57 GW of pumped-storage capacity by 2031-32

alongside substantial battery-storage deployment.

Storage allows excess renewable electricity generated during periods of high output to be used when production declines.

For AI data centres seeking low-carbon, round-the-clock power, storage could become a critical part of the energy architecture.

Nuclear Power Could Support AI Infrastructure

India is also examining nuclear power as a potential source of reliable low-carbon electricity for data centres.

Unlike solar and wind, nuclear plants can provide continuous power.

The government has specifically linked future deployment of:

small modular reactors

and

micro nuclear reactors

with emerging high-demand sectors such as AI and data centres.

Small modular reactors could eventually provide dedicated or near-dedicated clean electricity for large industrial and computing campuses.

However, commercial deployment at scale remains a longer-term opportunity.

Power Purchase Agreements Could Become Standard

Large data-centre operators increasingly secure electricity through long-term power-purchase agreements.

These arrangements allow companies to contract directly with renewable-energy producers.

Long-term contracts can provide:

price visibility,

renewable-energy access,

and support for new generation projects.

India's Green Energy Open Access Rules provide another mechanism for large consumers to procure renewable electricity.

As AI campuses become larger, developers may increasingly secure power before finalising data-centre construction.

Data-Centre Developers Could Become Energy Investors

The scale of future AI infrastructure may blur the boundary between data-centre development and energy development.

A company constructing a multi-hundred-megawatt AI campus may need to invest directly or indirectly in:

renewable projects,

storage,

substations,

transmission,

and backup systems.

Power procurement could therefore become a core strategic capability for data-centre operators.

Developers with strong energy partnerships may gain a major competitive advantage.

Cooling Creates Another Infrastructure Challenge

Electricity is not the only resource affected by AI infrastructure.

GPU servers generate substantial heat.

As rack densities rise, traditional air cooling becomes less effective.

The industry is therefore adopting technologies including:

direct-to-chip liquid cooling,

immersion cooling,

adiabatic cooling,

and closed-loop liquid cooling.

These systems can improve cooling efficiency for high-density AI workloads.

They can also influence water consumption and overall facility design.

High-Density AI Racks Change Data-Centre Engineering

Traditional enterprise servers may consume relatively modest amounts of electricity per rack.

AI infrastructure can operate at dramatically higher densities.

A single rack containing advanced GPU systems can consume tens or even hundreds of kilowatts.

That creates engineering challenges involving:

power distribution,

cooling,

floor design,

networking,

and backup infrastructure.

Existing data centres designed for conventional cloud workloads may therefore require substantial upgrades before they can host the newest AI systems efficiently.

Water Availability Will Also Matter

Some data-centre cooling technologies consume significant amounts of water.

That creates potential challenges in water-stressed regions.

India's data-centre expansion therefore needs to consider both:

energy efficiency

and

water efficiency.

Closed-loop cooling, direct liquid cooling and other advanced technologies can reduce water requirements.

The Bureau of Indian Standards has developed metrics covering:

Power Usage Effectiveness,

Carbon Usage Effectiveness,

Cooling Efficiency Ratio,

and Water Usage Effectiveness.

These standards could become increasingly important as the industry scales.

Energy Efficiency Could Reduce Infrastructure Pressure

Not every solution requires building additional generation.

Improving data-centre efficiency can reduce the amount of electricity required for each unit of computing.

Power Usage Effectiveness, or PUE, measures how efficiently a data centre uses electricity.

A facility with a lower PUE directs a greater proportion of its electricity toward computing rather than:

cooling,

lighting,

power conversion,

and other supporting systems.

As India's AI infrastructure grows, improvements in PUE could save significant amounts of electricity.

Semiconductor Efficiency Will Also Matter

The energy challenge extends all the way to the chips inside the data centre.

More efficient AI accelerators can perform greater amounts of computation using the same quantity of electricity.

This means semiconductor innovation could have direct implications for national energy requirements.

Future competition among AI-chip companies will increasingly focus on:

performance per watt,

memory efficiency,

cooling requirements,

and total cost of ownership.

For data-centre operators, electricity efficiency translates directly into operating economics.

India's Large Power System Is an Advantage

Despite the challenges, India enters the AI data-centre boom with a substantial electricity system.

The country has more than 500 GW of installed generation capacity, with non-fossil sources accounting for more than half of installed capacity.

Data centres currently represent less than 1% of national installed power capacity.

This means the problem is not an immediate nationwide shortage caused by data centres.

The challenge is planning ahead for extraordinarily rapid growth and ensuring power reaches the right locations with sufficient reliability.

India Could Benefit From Constraints in Other Markets

Power constraints are becoming a global data-centre issue.

Several established data-centre markets face difficulties obtaining:

new grid connections,

sufficient transmission capacity,

land,

and reliable electricity.

Some developers face multi-year waits for large power connections.

This creates an opportunity for countries capable of building infrastructure more quickly.

India could attract international AI workloads if it can combine:

competitive electricity,

large-scale land availability,

high-speed connectivity,

renewable energy,

and favourable policy.

Sovereign Cloud Adds Another Demand Layer

Sovereign cloud infrastructure is expanding alongside sovereign AI.

Governments and regulated industries increasingly want sensitive information hosted under domestic legal and operational control.

Potential customers include:

government departments,

banks,

defence organisations,

healthcare institutions,

telecommunications companies,

and critical infrastructure operators.

These organisations may require specialised domestic cloud environments rather than conventional global public-cloud infrastructure.

That creates another layer of demand for Indian data centres.

Data Localisation Supports Domestic Capacity

India's growing digital economy generates enormous quantities of data.

Regulatory and commercial preferences increasingly encourage certain information to be processed or stored domestically.

This strengthens the economics of local data-centre construction.

When combined with AI, the effect becomes even stronger.

Data used to train or operate AI models may need to remain within controlled environments.

Domestic compute infrastructure therefore becomes part of both technology strategy and data governance.

Data Centres Could Become Major Infrastructure Asset Class

The amount of capital required for the next generation of AI data centres is likely to be enormous.

Investors increasingly view data centres as infrastructure assets comparable with:

telecommunications towers,

renewable-energy projects,

logistics parks,

and utility infrastructure.

Large facilities require long-term capital but can potentially generate predictable revenue through multi-year customer contracts.

The AI boom could therefore attract capital from:

private equity,

infrastructure funds,

sovereign wealth funds,

pension funds,

and institutional investors.

Power Availability Could Influence Data-Centre Valuations

As electricity becomes scarcer in high-demand locations, access to secured power could itself become a valuable asset.

Two identical land parcels may have dramatically different commercial value if one has a confirmed 200 MW grid connection and the other does not.

This could change how investors evaluate data-centre projects.

Future valuations may increasingly depend on:

secured power capacity,

renewable-energy arrangements,

grid connectivity,

water availability,

and expansion rights.

Land alone will not be enough.

AI Infrastructure Could Support India's Digital Exports

If India successfully solves its infrastructure challenges, AI data centres could create a new category of digital exports.

Indian facilities could provide computing services to companies located internationally.

Instead of exporting only:

software,

IT services,

and business-process services,

India could increasingly export:

compute.

International customers could rent GPU infrastructure located inside Indian data centres.

This would create a new layer within India's technology-services economy.

Economic Benefits Extend Beyond Data Centres

Large AI campuses create demand across numerous supporting industries.

These include:

construction,

electrical equipment,

transformers,

switchgear,

cooling systems,

renewable energy,

battery storage,

network equipment,

fibre infrastructure,

engineering,

security,

and facility management.

The data-centre boom could therefore generate substantial industrial investment beyond the technology sector itself.

India's domestic electrical-equipment manufacturers could become major beneficiaries.

Execution Will Determine Whether India Captures the Opportunity

Demand for AI computing is unlikely to be India's primary problem.

The country has:

a large digital economy,

rapid AI adoption,

government support,

international investment,

and a growing startup ecosystem.

The challenge is execution.

India must simultaneously expand:

data-centre capacity,

electricity generation,

transmission infrastructure,

renewable power,

energy storage,

water-efficient cooling,

and high-speed connectivity.

Failure in any one of these areas could slow the broader ecosystem.

Power Strategy Could Decide India's AI Infrastructure Position

The global AI infrastructure race is increasingly becoming an energy race.

Advanced GPUs can be purchased.

Data-centre buildings can be constructed.

Capital can be raised.

But computing capacity cannot operate without reliable electricity.

Countries capable of delivering abundant, affordable and increasingly clean power will therefore have a structural advantage.

For India, this makes energy policy an important part of AI policy.

The country's ability to coordinate the two could determine how much of the global AI infrastructure boom it ultimately captures.

Conclusion

India has the demand, digital scale, technology talent and investment momentum required to become a major global AI data-centre hub, but electricity availability could become the decisive constraint on that opportunity.

Domestic data-centre capacity has already expanded from approximately 375 MW in 2020 to around 1,575 MW, while industry projections suggest operational capacity could eventually reach 12 GW by 2030.

At the same time, government estimates point toward a dramatic increase in electricity requirements as AI workloads scale.

Sovereign AI, sovereign cloud, data localisation, the IndiaAI Mission and international demand for GPU computing could push billions of dollars into Indian digital infrastructure.

But the next generation of AI campuses will require much more than servers and buildings.

They will need:

reliable grid connections,

renewable generation,

energy storage,

transmission capacity,

efficient cooling,

and long-term power planning.

India's opportunity therefore sits at the intersection of two major transformations:

the AI revolution and the energy transition.

If the country can expand computing infrastructure and electricity systems together, it could emerge not only as one of the world's largest consumers of AI but also as a major global provider of AI computing capacity.

If power infrastructure fails to keep pace, electricity could become the bottleneck that limits one of India's largest emerging digital-infrastructure opportunities.