Hiranandani said India is catching up on conventional cloud capacity, which currently stands at about 1.5 gigawatts. He said this needs to rise to at least three gigawatts by 2030, even before accounting for AI workloads. “To go from 1.5 gigawatts to 3 gigawatts, we require around $40 billion,” he said, including buildings, power systems and servers.
He added that if India moves towards 5 gigawatts of capacity by the end of the decade, total capital expenditure could rise to $70–80 billion. “Infrastructure can’t be built overnight,” Hiranandani said, noting that power, servers and cloud systems take years to deploy. He expects demand to remain visible through 2030, with limited risk of overcapacity in India due to the low starting base.
Also Read | Anant Raj to invest ₹4,500 crore in Andhra Pradesh for new data centre, IT park
Sarin said real estate is only the first step in the data centre value chain, with long-term returns coming from operating models such as co-location and cloud services. “The first thing you need is land and a building, but that is just the starting point,” he said.
Anant Raj currently has 157 megawatts of data centre buildings ready and about 28 megawatts of operational co-location capacity. Sarin said the company expects capacity to rise to 63 megawatts next year and 117 megawatts thereafter. “The roadmap for the next three to four years is very clear,” he said.
On returns, Sarin said co-location typically offers a return on investment (RoI) of 60–70%, while cloud infrastructure services can deliver higher returns once scale is achieved. Hiranandani, however, said payback periods vary. “Co-location paybacks are around six to seven years,” he said, while cloud compute investments can recover costs in three to four years due to faster depreciation cycles.
Also Read | OpenAI, NextDC plan to build $4.6 billion Sydney data centre
Both executives said AI becoming cheaper would not reduce demand for data centres. “If AI becomes cheaper, data consumption goes up,” Sarin said. Hiranandani added that even without global AI demand, India’s domestic cloud needs alone support several years of capacity expansion.
They also pointed to opportunities across the broader ecosystem, including power systems, mechanical, electrical and plumbing (MEP) suppliers and specialised data centre infrastructure providers, as capital spending gathers pace.
For the full interview, watch the accompanying video
Catch all the latest updates from the stock market here
