Data centre and AI cloud firm Yotta Data Services plans to deploy about $7 billion in capital expenditure this year and double that amount next year as it expands its GPU cloud business and positions itself as a global provider of AI infrastructure services, a senior company executive said.
“Yotta has already deployed $3 billion in capex this year and expects another $4 billion or more to be invested between November and January. The investments will help raise its operating capacity to around 175-180 megawatts by the end of the year, primarily supporting its cloud and AI services business,” said Darshan Hiranandani, chairman and co-founder of Yotta Data Services.
Hiranandani said the company has the ability to scale significantly within its existing campuses. Its Mumbai facility has been designed for up to 2 gigawatts of capacity, while the Greater Noida campus can support another 250 megawatts.
The company is also evaluating various capital-raising options, including a domestic listing, although it has not set a timeline. It plans to transition into a public limited company and begin making broader public disclosures even before any eventual listing.
The company, which operates data centre campuses in Mumbai and Greater Noida, expects 75 per cent of its revenue this year to come from global customers, underscoring its strategy of serving international demand from India rather than focusing solely on the domestic market.
“We are not a company that wants to deliver goods and services only to Indian customers. This year, 75 per cent of our revenue will come from global companies,” the executive said in an interaction.
The aggressive expansion comes amid rising global demand for AI computing infrastructure. Yotta said it has already contracted more than 30,000 additional graphics processing units (GPUs) and has room to deploy another 45,000 GPUs in the future.
The company was among the first in India to bring large-scale Nvidia GPU clusters into operation and today serves global customers seeking AI training and inference capabilities.
According to Hiranandani, the challenge in building an AI cloud business extends far beyond acquiring GPUs. “Buying GPUs is the easiest part. The difficult part is running the clusters, configuring applications and learning how to operate at scale,” he said.
Yotta’s AI cloud ambitions have also benefited from recent geopolitical developments. The company said some AI workloads originally planned for West Asia have been temporarily redirected to other regions, including India, amid uncertainty in the region.
“Some of the AI deployments that were going to go there have gone to other parts of the world, including to us. On a temporary basis, we have benefited from that shift,” he said.
The company also said it is seeing growing international interest in the sovereign cloud infrastructure it has built for India, with requests to replicate similar solutions in overseas markets.
While Yotta continues to provide colocation services, including being one of the largest providers of data centre space to Google in India, cloud-managed services, GPU cloud and AI services now form the core of its growth strategy.
Unlike many domestic data centre operators that focus primarily on colocation, Yotta sees itself as competing in the global AI infrastructure market.
“We cater to European, Singaporean and American customers who compare us with providers around the world. We are focused on global demand and global customers,” said Hiranandani.
