Nvidia partner DDN bets on sovereign AI boom as India builds massive data infrastructure

India’s push to build sovereign artificial intelligence (AI) infrastructure is emerging as one of the biggest drivers of global demand for high-performance data platforms, according to DataDirect Networks Inc. (DDN), whose technology underpins some of the world’s largest AI supercomputers.

In an interview with Business Today, Atul Vidwansa, Managing Director for India and Vice President for Worldwide NVIDIA Cloud Providers at DDN, said governments, not private companies, are increasingly the primary buyers of large-scale AI infrastructure.

“Sovereign AI is one of the biggest consumers of this technology,” Vidwansa said, noting that national security and technological independence are only part of the motivation. Countries are also seeking to build exportable AI capabilities. “These two drivers are pushing every nation to adopt AI… AI is a tool to move them ahead much faster than the traditional pace.” 

The comments come as India accelerates its IndiaAI Mission, aiming to develop domestic large language models, secure data infrastructure and national AI capacity. A major step in that effort was Yotta Data Services’ deployment of DDN storage systems to support thousands of NVIDIA B200 graphics processors inside its Shakti Cloud platform, one of the largest sovereign AI installations in Asia.

Vidwansa said Yotta, India’s largest NVIDIA Cloud Partner, is already serving government demand for such infrastructure. “Their biggest customer is the government of India,” he said, adding that the same pattern is visible globally. 

DDN’s systems are widely used across AI computing clusters because they manage the massive volumes of data needed to keep GPUs running efficiently. NVIDIA itself relies on the company’s technology internally for its own AI platforms.

The Yotta partnership reflects both technical capability and policy alignment, Vidwansa said. 

“One of the reasons Yotta decided to work with DDN is because we had the right feature functionality… tested and validated by NVIDIA,” he said. “At the same time, we had a made-in-India footprint that the government of India was looking favourably at.” 

He added that operating large AI clusters requires specialised expertise inherited from the supercomputing industry. “It’s very difficult to keep all these very expensive GPU clusters running 24/7. You need the right people for that,” he said. 

Beyond infrastructure, many organisations struggle to extract returns from AI investments because preparing data for training models is costly and time-consuming. Vidwansa said DDN has developed tools to automate this process.

“A huge portion of that pilot gets spent on making their data AI-ready,” he said. The company’s HyperPod system can automatically tag images, audio, and other files while they sit unused, reducing months of manual annotation work. 

India’s rapid build-out of data centres, essential for housing power-hungry AI chips, will be critical to sustaining the country’s ambitions, he said. Tax incentives are less important than the strategic necessity of the infrastructure itself.

“Data centres are becoming a necessary infrastructure for any nation’s AI capability development,” Vidwansa said. “If you don’t have data centres, you can’t deploy AI.” 

He compared the effort to building railways or highways, expensive but unavoidable. “This is not a luxury… We have to build this as a critical infrastructure required for a nation,” he said.