Nandan Nilekani doesn't think India's AI advantage will come from chips, infrastructure, or building foundation models. Speaking at the India Policy Forum, the Infosys co-founder and chairman argued India could become a principal beneficiary of AI-driven productivity gains over the next decade, despite limited presence in those upstream layers of the stack.

The applied-AI bet

Nilekani's argument is specific: India's edge is deployment, not development. "It's in the area of applied AI that India will become a leader," he said. "We will become the AI use case capital of the world because we will actually deploy AI and make it useful, not just for a few businesses or a few people, but make it accessible and useful to a billion Indians."

He compared this directly to India's earlier bet on digital public infrastructure — UPI, Aadhaar, the DPI stack — framing applied AI at population scale as the next run of the same playbook: build for a billion people, not a premium segment.

An argument against conventional wisdom

Nilekani, who also chairs the think tank NCAER, made a claim that flips the usual assumption about who's safe from AI-driven job losses. Large, well-structured companies, he argued, are actually the most exposed, precisely because their organisation is mature enough that every role has already been broken into a defined set of tasks. That legibility is what makes automation easy.

Small businesses are better positioned. Many are effectively single-person operations already, doing work that doesn't decompose as cleanly into a task list a model can absorb.

The prediction follows: "the future will be small businesses," with growth coming not from a handful of large employers scaling up but from many more small, often single-person companies. Instead of one company hiring a million people, he expects a million companies with one person each.

Why the framing matters

Most commentary on India and AI defaults to the infrastructure gap — GPU access, compute costs, no homegrown frontier model. Nilekani is making a different argument: that for a market India's size, the constraint that matters isn't who builds the best model, it's who gets AI into the hands of the most people doing real work. That's testable, and India's own DPI track record gives it some grounding.

The jobs argument is more speculative and should be treated that way. It's a prediction about how automation interacts with organisational structure, not a documented trend. But it's a genuinely different frame from the standard "AI hits low-skill jobs hardest" story, and it comes from someone who has actually built population-scale infrastructure in India before.

Source: Times of India — India may see productivity gains from AI, says Nilekani