India as a Global Data Centre Hub Powered by Renewables

A roadmap to set-up, scale and sustain

Context and market dynamics

• Capacity gap -- While there are more than 40 GW datacentres in the U.S., India has only 1.6 GW currently
• Supply crunch -- There will be a load gap of more than 114 GW in the U.S. over the next seven years, even after assuming approximately 76 GW of capacity addition from gas-based power plants. This is on account of the U.S.’s total electricity demand expected growth of 17% CAGR from 2025 to 2030, primarily driven by datacentre and new manufacturing demand 

The India advantage

Given the above realities, datacentre expansion will have to evaluate other countries. With the geopolitical instability of the Middle East eroding global investor sentiment, India stands out as a premier contender for both scale and stability. Our recent milestones underscore this potential -- in 2025, India added the second highest renewable capacity in world (we overtook the U.S. in solar; and both US and Germany in wind capacity addition).

Our key differentiators --

• India’s grid can provide large-scale 100-200 MW connection to a datacentre within 2-3 years, whereas similar projects in the U.S. face a timeline of 4-5 years or more
• Electricity in India remains highly competitive at approximately $55/MWh, nearly half the $100/MWh levelized cost of U.S. gas-based power (CCGT)
• Besides the above, this electricity in India is largely supplied from renewable power (about 60-70%) as compared to the U.S. This shift is critical for decarbonization. It is critical to note that for every 10 GW of data center capacity, switching to India’s green-heavy mix will prevent roughly 0.13 billion tons of CO2 emissions

Which bring us to few key questions --

• What do Hyperscalers care about? Should a datacentre be located only near a Tier-1 or Tier -2 city or can it be located near a grid substation?
• Does India’s grid have the capacity to feed large scale datacentres? Will it be a reliable one? Should it be around central grid or state grid?
• Could India also build an off-grid solution for feeding datacentres? How will a renewable generator (solar and wind) feed reliable power to a datacentre? Can a datacentre get reliable power without gas-based / nuclear power plant?

When looking at these objectively, here is my take

Q 1: What do Hyperscalers care about?

A, That they need absolute computation capabilities as soon as possible (asap), which translates to a colossal amount of energy asap. And here is what is interesting – while they prefer cheap and clean energy first and foremost, their main lookout is scale and speed

B Datacentres for inference vs training have different energy requirement. Inference requires proximity to end users and a high degree of reliability and redundancy in energy supply. Training datacenters are more flexible geographically, and can tolerate less redundancy, though high GPU costs still penalize any significant downtime. This table summarizes the key requirements and prioritization:

Criteria Priority Requirement Scale P0 5 to 8 GW in next 5 years in India
Minimum size: 50 MW on contiguous land
Minimum cluster: 500 MW total capacity in 15 km Speed P0 While the U.S. views co-located natural gas as the fastest option when grid capacity is unavailable, in India one can set up a datacenter within a 100 to 200 km radius of a Tier 2 city and secure an STU connection within 2 years (especially if the organization lays the transmission line to the nearest substation). If setting up within a city for inference, the retail connection may take more time, specifically as we expect to get more demand in the future.

The preferred model in India will be a retail connection from utility, while procuring renewable power using open access, leading to lower cost of electricity and lesser emission intensity Reliability P1 Both power coming from two STU routes or a combination of CTU and STU connections can be considered reliable. A CTU connection, paired with a contracted PPA, should also be reliable enough. In all these cases, DC sets or BESS can serve as back-up power Cost P1 A datacenter in India can target power costs of <$55/MWh, compared to < $100/MWh in the U.S.

In India, setting up an off-grid datacenter using gas-based power plants could be as costly as $150/MWh. Furthermore, this tariff would fluctuate significantly based on prevailing gas prices. Emission Intensity P2 In India, a datacenter can easily replace 70% of grid power with renewables, leading to significant emission savings

Q 2: Does India’s grid have the capability to feed datacentres?

In India, central bodies like CEA (Central Electricity Authority, the technical wing of Ministry of Power) and CTU (Central Transmission Utility), plan transmission network additions based on estimated generation and consumption points rather than waiting for formal connectivity applications from generators or loads. CEA has already announced a comprehensive transmission plan to support integration of 900 GW of non-fossil fuel based power generation capacity by 2035-36. This plan includes building more than 1,37,500 kms of transmission lines and 8,27,600 MVA of transformation capacity in the next 8 years (up to 2035).

In India, in most of the cases, one can get a 100-200 MW retail connection within 2-4 years. If a datacentre is built near a STU substation, the STU connection along with dedicated line can be built within a year.

A datacentre in India has the following options for a power connection (other than off grid):

1. Get a connection from local distribution licencee (discom): In India, securing a large grid connection for a data centre typically requires a lead time of one to four years. However, if faster connectivity is a priority, a data centre operator can reduce this timeline by laying dedicated cables or overhead lines to the nearest STU substation, or by formally engaging the STU to complete the connection within an agreed schedule.
   As contracted demand from data centres grows, utilities are likely to face increasing pressure on delivery timelines, making alternative procurement routes a strategic consideration for the long term. It is also worth noting that Andhra Pradesh has recently moved to grant distribution licences directly to data centres, a policy development that other states are expected to evaluate and potentially replicate
2. Get a CTU connection (if the load is more than 50 MW, as is typical for a datacenter). At the CTU level, you may opt to sign a long-term power purchase agreement (PPA) with a generator, preferably one also connected to the CTU, to increase reliability. While it is rare to face a total lack of power from the exchange, spot market tariffs remain uncertain. Typically, a CTU connection offers greater regulatory certainty, as it is driven by central government policies which are considered more stable than state-level ones. A CTU connection backed by DG sets should be considered a reliable source; however, to further increase reliability, one may opt for an additional STU connection.

Q 3: Could India also build an off-grid solution for feeding datacentres?

India has a gas network and a decent 25 GW plus of gas-based power plants. However, hardly 8 GW of them are operational and even those run at a low CUF (Capacity Utilization Factor). Most of the investments in gas-based power plants in India have turned into non performing assets (NPAs) owing to unviable electricity tariff as high as 13-15 Rs/u (around $128/MWhr), and their high susceptibility to gas price fluctuations. A datacentre which can afford such steep electricity cost at the cost of stable power can try locating their datacentre near the LNG terminals of gas pipelines.

In the United States, the push towards off-grid power for data centres has been driven largely by the delays and constraints in securing grid connectivity. Against this, this is challenge is more manageable in India currently; 100-200 MW contract demand connectivity is easily available within one to four years. However, as data centre load growth compounds alongside rising demand from manufacturing plants and electric vehicles, operators will need to begin planning for off-grid solutions well ahead of that inflection point.

For India, a portfolio of solar, wind, BESS and DG sets can meet the requirement of off-grid solution, at rates much cheaper than gas-based power plants. Two more options are on the anvil: one, long duration batteries, anticipated to be commercially viable and mainstream in the near future, will offer a good option. Pumped hydro storage, option two, though technically sound, carries long gestation periods. Geological uncertainties make it a less reliable planning assumption though, thereby shifting preference to long duration battery storage (option a) instead.

India offers the perfect datacentre ecosystem, rates, and the promise

India offers a compelling and increasingly complete data centre ecosystem with competitive costs, abundant renewable resources, and a domestic demand story that is only getting stronger. Our solar and wind resource base is among the most advantageous in the world, and with land available at scale, renewable tariffs are already highly competitive. As battery storage costs continue to fall, the combination of solar, wind and BESS is fast becoming the most cost-effective and reliable power solution for data centres in India.

On transmission, significant network expansion is already underway, anticipating the country's growing electricity demand, and data centre loads. Our domestic demand case is equally strong. With 750 million internet users, India's AI-driven data consumption will be substantial. And as data localisation requirements tighten globally, with more countries mandating that certain categories of data be stored within national borders, India's case as a preferred data centre destination will only strengthen. The infrastructure investment thesis here is therefore structural, proactive, and already work-in-progress. India is deliberately building to carry the world’s AI load with clean, reliable, and certain energy.