NEW DELHI: India’s apex power sector planning body has gone back to the drawing board to map electricity demand amid concerns that a concentration of data centres coming up in a few states could skew consumption and adversely affect national grid stability.
The Central Electricity Authority (CEA) will also ask states and power distribution companies to incorporate the power demand projections of data centres in their resource adequacy plans, said two people aware of the developments. The hubs are expected to lead to a surge in electricity demand, significantly changing the demand profile of the host states and possibly impacting grid stability.
Adding the projected electricity consumption of data centres in the resource adequacy plans of states is aimed at ensuring there is sufficient , storage and transmission capacity to meet consumer demand.
The government expects to attract $200 billion of investments in data centres and infrastructure by 2030, with their capacity projected to surge from about 1.2GW in 2025 to 10GW and their electricity consumption estimated to reach 40-45 terawatt hours (TWh) by the end of the decade, up from 10-15 TWh consumed in 2024.
One terawatt hour is equivalent to using 1 trillion watts of electricity continuously for one hour. A standard LED bulb uses about 10 watts.
Large data centres are expected to come up in Andhra Pradesh, Uttar Pradesh, Maharashtra, Telangana, Karnataka and Tamil Nadu. Companies that have committed to invest in India’s AI space include Microsoft, Amazon, Alphabet, the Adani Group, Reliance Industries, the AM Group and Larsen and Toubro.
“The CEA is working on demand projections for data centres and will add the requirement for estimation of data centre demand in resource adequacy plans,” one of the two people said.
Demand assessment
A resource adequacy plan estimates the quantum and type of energy resources required to meet power demand at the least possible cost. The plan assesses annual power generation capacity, including from conventional and renewable sources, storage capacities, reserve margins and costs. According to the guidelines for resource adequacy planning, the exercise must have a horizon of 10 years on a rolling basis.
Spokespersons of the power ministry and the CEA did not respond to queries emailed by Mint till press time.
The parliamentary standing committee on energy, which submitted its report to the Lok Sabha last month, noted that India’s current data centre load of about 1.2GW is projected to increase to 13.6GW by 2032 and 16.4GW by 2040.
The power ministry informed the parliamentary panel that an increase in electricity demand on account of rapid growth of AI and other large-scale data centres and other emerging loads is being ascertained under the 21st Electric Power Survey Exercise. The increase in demand will be considered in the National Electricity Plan for the period from FY28 to FY37, which is expected to be released by March 2027, the ministry said.
According to Debasish Mishra, chief growth officer at Deloitte South Asia, data centres or clusters are likely to consume about 2GW of power each around the clock, which will be added to the base load. They have the potential to change the demand profile of the area and the state, which can impact the national grid, Mishra said.
For Maharashtra, Andhra Pradesh and Karnataka, 2-3GWwill make a huge difference, for which planning will be required. Instances of grid instability due to sudden surges in demand may lead to a grid collapse, he said.
No major impact
However, Alok Kumar, former power secretary and current director general of the All India Discoms Association, noted that data centre hubs may not have a major impact on the country’s peak demand load.
“Data centre plans and projections have been very ambitious across the globe,” Kumar said. “They have not materialized as expected even in major economies. Maybe 5 to 8GW of cumulative data centre capacity will be installed by 2030.”
Kumar said there could be other causes of concern in emerging areas such as space cooling including households.