AI predicted to take India forward
Tapadia said, from an AI workload perspective, although both global hyperscalers and Indian startups have started considering India as the new training ground for large language models (LLM), the next phase of growth will continue to be driven by AI inference companies.
“Globally, we do not build gigawatt-scale campuses. Our sweet spot is building data centers with 100-500 megawatt (MW) capacity, and that is where we are primarily positioning ourselves. In India, the projected workload growth will be much higher,” he said, adding that since the company’s campuses will have enough capacity for both, Iron Mountain will be able to meet some of the domestic demand from Indian startups working towards LLM training.
The company plans to add 150 MW of data center capacity in India over the next three years, with three campuses currently under construction in Mumbai, Chennai and Noida. He said these complexes are planned to have capacities of 85 MW, 50 MW and 15 MW respectively and will be operational by 2028.
With increasing AI workloads, the demand for data centers is also changing, Tapadia said, adding that most of the older data centers are now being retrofitted with new capacity servers, racks and cooling infrastructure to handle the increased workloads.
“We don’t have that much legacy in India, but that’s an advantage for us. All our buildings are new and AI-ready, which mainly means that, depending on how the facility is built, we can quickly transform the entire facility into an AI-native facility,” Tapadia said.
He said, one of the other advantages of data center business in India is the surplus availability of stable power along with green power generated from renewable energy sources like wind, solar and hydropower.
“India has all the ingredients to become a regional hub for data centres, and there are multiple drivers for it. Obviously, the domestic market is booming with rapid digitalization and increasing cloud adoption. AI is driving overall demand,” Tapadia said.
