India revives national datacentre policy amid AI push

Policy priorities

The policy refresh aims to replace more than 30 permissions needed for datacentre builds with a single-window clearance system, cutting delays by months, operators said. It is also planning four datacentre economic zones (DCEZs) to drive datacentre expansion into smaller cities, easing concentration in Mumbai, Chennai and the Delhi National Capital Region (NCR), which together account for nearly 80% of India’s datacentre capacity.

Operators see these reforms as critical. “AI, quantum and post-cloud architectures need faster, safer and more capable datacentres. With the right blend of talent, policy and vision, India can tap opportunities nationwide,” said Anil Pandey, managing director of Kennies Data Center, in a recent report.

The government is trying to draw from more than 10 state policies in its national policy, but integrating them is a major challenge. Industry association Nasscom has warned that the fragmented policy landscape is already forcing companies to deal with a prolonged approval process that drives up costs. Fast-tracked clearances in Maharashtra, for example, contrast sharply with delays from stricter land rules in Tamil Nadu.

Analysts noted that India’s national datacentre policy would need to address infrastructure challenges to make datacentre expansion smoother. “States offer subsidies, but growth beyond top cities needs robust fibre networks, reliable power and local talent,” said NSN Murty, partner and leader for government and public services at Deloitte India.

Arun U, principal analyst at QKS Group, pointed out the challenge of meeting the power requirements of AI workloads. “Datacentres require 99.995% uptime, but many regions face grid disturbances,” he said. “AI racks now draw 50–200 kW, pushing operators towards liquid cooling, and high-density power setups are still rare in India.”

Naresh Singh, senior director analyst at Gartner, added that the national plan “should allow operators to run captive power plants using renewable, natural gas, or even nuclear energy to meet future demand”.

The AI boom is making these challenges urgent. The IndiaAI Mission’s expanded 34,000 compute capacity will depend on decentralised infrastructure, without which, AI adoption in sectors such as financial services and manufacturing could face major bottlenecks. MeitY has cited India’s first AI datacentre park in Naya Raipur, Chhattisgarh, which opened in May 2025, as a model for decentralisation.

For CIOs and cloud strategists, India’s appeal lies in its engineering talent, cost advantages and proximity to Asian markets. But as Arun warned, “without harmonised reforms, lower tariffs and faster grid upgrades, India risks falling short of its AI ambitions”.

Consultations will run through August 2025, with the final policy expected later this year. The government’s ability to cut red tape, expand capacity and enable next-generation workloads will ultimately shape India’s position in the global AI and data infrastructure race.