Infosys Co-founder and Non-Executive Chairman Nandan Nilekani cautioned against overestimating AI-driven productivity gains based on greenfield development, arguing that the real enterprise challenge lies in modernising decades-old, complex legacy systems.
“Writing greenfield is not a big deal. I can take a tool, give it to a kid, and he'll generate a million lines of code. But that's not the real world. The real world is the fact that companies have trillions of dollars invested in the [legacy] systems. They have technical debt, they have data silos, they don't have documents.”
Speaking at Infosys Investor AI Day 2026, he pointed to the structural complexity inside large enterprises, including undocumented dependencies and legacy systems that few understand.
“Taking brownfield systems and modernising them is… a lot more difficult than doing greenfield development,” he explained.
In some cases, he noted, companies still rely on veteran specialists, who must now be in their seventies, to troubleshoot older systems because institutional knowledge was never fully documented.
Nilekani made the comments at a time when the Nifty IT index saw historic lows, triggered by a mix of technological and macroeconomic factors. Anthropic recently rolled out upgrades and enterprise capabilities for its Claude AI models that can automate routine enterprise tasks, unsettling IT services companies dependent on US technology spending.
Problem is deployment, not models
Addressing concerns around rapid advances in frontier AI models, Nilekani said enterprise constraints—not model capability—are the real bottleneck. “It's about how fast companies can implement it.”
He described what he called a “deployment gap” between the pace of technological progress and the slower AI implementation at scale due to organisational complexity, retraining needs, and legacy integration.
“It is not an opportunity risk, it's an execution risk,” he noted.
Talent Transformation is “Huge”
Nilekani said the AI shift will reshape workforce roles rather than eliminate the need for talent.
“Now I think talent transformation is huge. It's not that you won’t need talent, but it will go [away] from QA, testing or development. We have all kinds of new roles—AI engineers, forward deployment engineers, AI leads, forensic analysts, data [scientists],” he noted.
Hiring, training, and deployment models will change, he said, as writing code becomes less central and orchestration and AI engineering gain importance.
Legacy Modernisation Cannot Be Deferred
Over decades, enterprises have layered technologies—mainframes, minicomputers, LAN systems, and cloud—without fully replacing older systems. Many companies now spend 60–80% of their IT budgets on maintenance.
Nilekani said accumulated technical debt “must be paid” and that firms no longer have the option to defer modernisation if they want to benefit from AI. Security risks and siloed data further add to the urgency.
Your reaction
Discussion
What's your take on this story?
No comments yet — be the first to share your view.