In the global race for artificial intelligence supremacy, Nandan Nilekani, co-founder of Infosys and architect of India’s landmark digital public infrastructure projects like Aadhaar and UPI, offers a compelling counter-narrative. Rather than competing head-on with Silicon Valley in building ever-larger frontier models, India should focus on rapid diffusion and large-scale, practical applications. This approach, he argues, could deliver greater real-world impact and position India ahead in the long term.
The Two Parallel Races in AI
Nilekani distinguishes between two distinct competitions unfolding in AI development. The first is what he calls the “Race to the Bottom”—Silicon Valley’s intense focus on scaling massive large language models (LLMs) with hundreds of billions in investment. While this drives technological breakthroughs, it also amplifies challenges: potential mass displacement of knowledge workers, enormous energy and water consumption by data centers, and rising societal concerns over addictive interfaces and misinformation.
The second, more meaningful contest is the “Race to the Top”—leveraging AI to tackle pressing societal problems such as climate change, healthcare access, agricultural productivity, education, and drug discovery. Victory here depends not on raw model size but on diffusion: making trustworthy, localized AI tools accessible to millions of people, especially in diverse and resource-constrained environments.
India, according to Nilekani, is uniquely positioned to win this second race by building on its proven strengths in scale, linguistic diversity, and inclusive digital infrastructure.
Real-World Success Stories from India
Several early deployments illustrate this strategy in action. One standout example is Sarla Ben, an AI assistant developed for the Amul dairy cooperative. Built in roughly three weeks, it serves over 2.6 million farmers managing 22 million cattle with a network of just 1,400 veterinarians. Farmers can query the system in Gujarati about livestock health, receiving practical advice that directly improves incomes and productivity.
Similarly, Maha Vistar, Maharashtra’s multilingual agriculture bot, delivers tailored guidance on crops, soil, seeds, and irrigation in local languages such as Marathi. Early lessons from this project have accelerated deployments elsewhere, including an export of the solution to Ethiopia within three months.
These initiatives are deliberately voice-first and designed for feature phones, ensuring accessibility for millions who lack smartphones. They build upon the trust and verification layers established by systems like Aadhaar and UPI, emphasizing that technology constitutes only about 30% of the solution. The remaining 70% involves data quality, stakeholder coordination, organizational change, and mechanisms to minimize AI hallucinations.
Why Frontier Models Are Becoming Secondary
Nilekani notes that the world already has around ten capable frontier models—primarily from the US and China—along with strong Indian contenders like those from Sarvam AI. As open-source alternatives proliferate, models are rapidly commoditizing. For most societal applications, current capabilities already exceed requirements.
The real bottleneck is not intelligence but deployment at population scale. Switching between models is relatively easy; integrating them into workflows for millions of farmers, students, or patients is far harder. Nilekani’s advice to Indian companies and policymakers is clear: let Silicon Valley invest in ever-larger LLMs. India should lead in use-case innovation, synthetic data generation, vertical-specific models, and trustworthy, multilingual deployment at national scale.
A Vision Grounded in India’s Digital Playbook
This AI strategy extends India’s successful digital infrastructure model. Just as Aadhaar provided identity to over a billion people and UPI revolutionized payments, AI can create inclusive, human-centered systems that modernize key sectors of the economy. By prioritizing aspiration, reskilling, and practical outcomes, India can turn its vast problems into opportunities for innovation—and potentially export these solutions to other developing nations.
Challenges remain, including limited resources for training top-tier models and the need to manage job transitions in the IT and services sectors. Yet Nilekani views AI primarily as a tool for modernization rather than disruption, provided India maintains focus on execution, trust, and inclusion.
Delivered during a session at the Raisina Dialogue 2026, Nilekani’s perspective underscores a fundamental truth in the AI era: the ultimate winners will not necessarily be those with the biggest models, but those who make AI most useful to the greatest number of people. India’s edge lies in its unmatched scale of real-world problems, rich data diversity, and proven ability to build systems that reach everyone.
About The Author
