Power grid, digital connectivity key hurdles in India’s AI push, say experts

Mumbai: As India pours billions of dollars into new artificial intelligence (AI) data centres, the country’s biggest challenge is no longer generating enough electricity but ensuring reliable power reaches these facilities through stronger transmission networks while building the digital infrastructure needed to distribute AI workloads efficiently.

Speaking at the Mint Sustainability Impact Summit 2026, held recently, industry executives and a policy expert said India’s AI ambitions will increasingly depend on upgrading electricity transmission, expanding fibre connectivity and planning data centre locations more strategically.

Sunil Gupta, chief executive of Yotta Data Services, said India already has adequate electricity generation capacity, and the real bottleneck lies in the last-mile transmission and distribution network.

“India is actually creating more power, generating more power than what India needs, except for peak months. The issue is not in terms of the overall capacity at the national level. The issue is at the distribution level,” Gupta said. He added that developers should shoulder a larger share of the upfront investment required for transmission infrastructure rather than spreading those costs across the broader electricity network.

Power generation, executives said, is only one part of the equation. The bigger challenge is building enough transmission infrastructure to move electricity from where it is produced to where AI campuses are being built.

Amit Paithankar, chief executive of Ergoniq, a deep energy intelligence platform, said India’s grid will need substantial upgrades as renewable energy accounts for a larger share of electricity generation.

“We are progressing, we are seeing that the amount of renewable on our grid is increasing. As the amount of variable renewable increases, people are saying instability has gone up. Renewable power does not have inertia. If something goes down, it just switches off, whereas there is a mechanical inertia associated with generators,” Paithankar said.

“The second biggest issue is that our transmission infrastructure itself is very, very low. So you need to put in more towers, more cables, more transformers. We just don’t have it.”

He said investments in battery energy storage systems, stronger evacuation infrastructure and grid upgrades would be necessary to maintain reliability as AI demand and renewable energy capacity continue to grow.

Strengthening the electricity grid alone will not solve the problem, according to Amajit Gupta, chief executive and managing director of Lightstorm Group, a digital infrastructure platform. He said India also needs a nationwide digital connectivity backbone capable of moving computing workloads across the country.

“About $50 billion of data centres are getting built over the next two years. Eighty per cent of them are built out in corridors, which are Hyderabad, Chennai, and Mumbai. The grid in this part will see an extraordinary explosion of power as well as water,” he said.

Lightstorm’s Gupta argued that concentrating AI infrastructure in a handful of cities would place disproportionate pressure on local electricity and water resources, even though advances in fibre connectivity now allow computing to be distributed much more widely.

“It is a myth that data centres should be near the sea. India has a long coastline. There is enough great-quality fibre being built and you could build your data centre and connect to the internet, including away from the seas,” he said.

Dispersing data centres across different regions would spread electricity demand, water consumption and economic activity more evenly instead of burdening a few existing hubs, he added.

Archana Tewary, partner at JSA Advocates & Solicitors, said future AI infrastructure should be planned around access to power, resources and skilled talent rather than continuing to cluster in established data centre markets.

“It is very important to site your AI and data centres in a dispersed manner. You need to have your data centres in places where power is accessible, resources are accessible and you have adequate talent,” she said.

The rapid growth of AI infrastructure has also intensified scrutiny of the sector’s environmental footprint, particularly its water consumption. Yotta’s Gupta, however, said comparisons with overseas facilities often overlook how India’s industry evolved.

“Globally, you are using huge amounts of water, but in India, because the industry came much later, the entire industry is using what we call a closed-loop system of water cooling and not an open loop,” he said.

Unlike many older international facilities that relied on once-through cooling systems, Indian data centres predominantly use air-cooled chillers and recirculate water through closed-loop systems, significantly reducing overall water consumption.

As AI capacity expands, executives said greater transparency around electricity and water use could help improve public understanding of the industry’s environmental impact.

Tewary noted that while governments encourage green and sustainable data centres through incentives, companies are not currently required to disclose AI- or data centre-specific resource consumption.

“When we ask for data centres to open in India, there are regulations that incentivise sustainable data centres and green data centres, but those are the carrots at the moment. There is no stick at present,” she said. “There’s no mandatory disclosure. Listed companies disclose their energy efficiency and ESG metrics, but those are not bifurcated to show how much goes into AI and data centre consumption.”

Yotta’s Gupta said the industry would support greater disclosure, provided reporting focuses on efficiency rather than absolute consumption.

“I think we should be reporting our power usage effectiveness, how much power is going towards cooling versus the power going to the equipment. We should be talking about water usage,” he said, arguing that metrics such as power usage effectiveness and water usage effectiveness provide a more meaningful picture of sustainability than total electricity or water consumption alone.