India is on the cusp of becoming a dominant force in the global artificial intelligence landscape, not by building the most sophisticated models initially, but by focusing on scalable and cost-effective solutions. This is the vision articulated by Pratyush Kumar, co-founder and CEO of the homegrown AI startup Sarvam, who believes India offers a unique opportunity to the worldwide AI effort.
India's Unique Cost Advantage in AI
Speaking at the Global Fintech Fest, Kumar emphasized India's distinct ability to develop AI at a cost structure that is viable for a population of over a billion people. He drew a compelling parallel between the burgeoning AI sector and India's remarkable journey in digital payments. Just a decade ago, India wasn't a global leader in digital transactions; today, it accounts for over 60% of the world's digital payment volume. Kumar anticipates a similar trajectory for AI.
“It's year four for the ChatGPT launch. But in some sense, it's year one for India building its own AI,” Kumar noted, highlighting recent advancements like the AI Impact Summit in Delhi, the emergence of domestic AI models, increasing GPU availability, and a rapidly upskilling talent pool. He firmly believes that within the next decade, India will lead in both AI adoption and development.
From 'Token Factories' to Scalable Infrastructure
Kumar stressed the necessity for India to produce intelligence at scale, likening it to manufacturing cars or steel. He introduced the concept of “token factories”—large clusters of GPUs dedicated to processing and generating AI outputs. Sarvam, under his leadership, has already demonstrated this capability, serving models at prices competitive with or even lower than global standards.
Sustaining this momentum requires robust infrastructure capable of both serving AI models and hosting comprehensive stacks for custom training. Sarvam's operational achievements underscore this potential: over the past year, the company has logged 325 million minutes of native voice AI, managed nearly 400 million API calls daily, and digitized millions of documents, collaborating with significant public sector institutions like NPCI and SEBI, as well as private banks.
Addressing Adoption Challenges and Fostering an Ecosystem
Despite these advancements, Kumar acknowledged that AI adoption in India remains somewhat fragmented, often operating in isolated silos. He pointed to existing apprehension and perceived risks associated with deploying AI solutions, whether for voice interfaces or document processing, presenting a range of challenges that need to be overcome.
To accelerate and secure AI deployment within regulated sectors, Kumar urged the industry to adopt an “ecosystem view.” This collaborative approach is crucial for ensuring that advanced technology effectively reaches tier 3, 4, and 5 regions, thereby promoting greater financial inclusion across the country.