Bengaluru, Jan 28 (.) India is set to become a global leader in population-scale voice AI, with far-reaching implications for language technology, governance, and everyday life, according to Nandan Nilekani, co-founder of Infosys and architect of India’s digital infrastructure.
Speaking at Voice AI – Making the Best Work for India, a technology event hosted by EkStep Foundation with NVIDIA as the technology partner on Wednesday, Nilekani said that successfully delivering voice AI across Indian languages over the next few years could create a playbook for the rest of the world.
“Globally, voice AI is essential. When we do something here, we create a model that other languages and regions can adopt,” Nilekani said. He added that while there are multiple AI models available, all are powered by Nvidia, providing the necessary infrastructure at scale. “Vishal Dhupar (NVidia South Asia Managing Director) is a lucky fellow. He just issues the orders: ‘I want a million GPUs.’ And it happens,” he quipped.
Nilekani highlighted the dual nature of AI development, describing a race to the bottom, involving low-quality content, pornography, and tools exploiting mental health and loneliness, versus a race to the top, focused on improving people’s lives. “We want India to lead the race to the top,” he said.
He cited India’s existing population-scale infrastructure as a model. “Aadhaar covers 1.3 billion people with the lowest authentication and KYC costs in the world, enabling banking and financial services. UPI handles 20 billion transactions at zero cost, so even a vegetable vendor selling ₹5 can get paid without fees. We’ve done this before, and we can do it again,” he said.
On India’s progress in AI, Nilekani said the country began supporting multiple languages five years ago, initially focused on translation. “Now it’s about voice, which is a whole new frontier due to variations, dialects, and code-switching between languages,” he said.
He emphasised the need for vertical-specific knowledge: railways for PNRs and bookings; judicial systems for court matters; agriculture for farmers, soil, and pricing; and education for how children learn. “We’re working on multiple languages, multiple dialects, and multiple verticals simultaneously,” Nilekani added.
On responsible AI use, he said guardrails are critical. “It’s about ensuring AI behaves as expected, doesn’t hallucinate, and stays within boundaries. Collaboration across governments, academic institutions, and industry is essential,” he said.
Nilekani shared his guiding principle: “My compass is utility. Is it useful? Relevant? Does it improve lives—education, healthcare, legal services, access to food, better prices? AI should amplify humans, not replace them. Starting from that perspective ensures the technology truly benefits people.”
. . CDS