Nvidia has recently launched a small Hindi-language AI model dubbed Nemotron-4-Mini-Hindi-4B. Made on 4 billion parameters, the AI model builds on CEO Jensen Huang’s view that each country should develop AI using its own infrastructure, local data, and workforce to ensure inclusivity.

Nvidia said Indian IT services firm Tech Mahindra is the first to use its new Hindi-language AI model to develop a custom AI model called Indus 2.0, focused on the language spoken nationwide and its dozens of dialects.

Further, Nvidia has stated that it will supply artificial intelligence processors to Indian companies such as Reliance Industries as it deepens partnerships to utilize a growing market. Notably, the U.S. company will supply its Blackwell AI processors for a one-gigawatt data center Reliance is building in Gujarat.

Nvidia also revealed its plans to supply tens of thousands of its Hopper AI chips to build large-scale data centers, in an expansion led by firms such as data center provider Yotta Data Services and Tata Communications.

“In the future, India is going to be the country that will export AI,” Huang said, in contrast with its role in software exports. “You have the fundamental ingredients—AI, data, and AI infrastructure—and you have a large population of users.”

Besides Tech Mahindra, Nvidia is partnering with IT giants such as Infosys, TCS, and Wipro to train about half a million developers to design and deploy AI agents using its software.