Nilekani said that adopting AI is relatively easier in the consumer space, but at an enterprise level, companies are putting their brand behind an AI-powered offering and even minor errors can affect their image
He further noted that one of the reasons enterprise AI is progressing slowly is the absence of guardrails to prevent machine errors in operations
Nilekani said that as the expectations from AI are huge, people don’t want even the slightest mistake
Infosys cofounder Nandan Nilekani believes that building enterprise AI at scale is challenging and takes time to mature, but the most difficult task is implementing AI in the public sector.
Speaking at the Carnegie Global Technology Summit 2025 yesterday, Nilekani said that adopting AI is relatively easier in the consumer space, but at an enterprise level, companies are putting their brand behind an AI-powered offering and even minor errors can affect their image.
He further noted that one of the reasons enterprise AI is progressing slowly is the absence of guardrails to prevent machine errors in operations.
“If the enterprise provides AI at scale and that AI even has 2% error in the way it answers, then that affects the brand itself. That’s one of the reasons why it’s taking so long, because we don’t have the guardrails to ensure that we have absolutely no machine error in these things,” Nilekani said.
He added that as the expectations from AI are huge, people don’t want even the slightest mistake.
While speaking about implementing AI in the public sector, he said that the public sector has structural consents, it has ministries, departments, everybody is territorial and data is not always shared.
“If data is the lifeblood of AI, we have to find a way to bring all of it together, irrespective of which part of the government it comes from,” he added.
He further underscored that the public sector requires clear commitments and no scope for blowbacks.
Nilekani in his address also stated that the hype levels for AI are unprecedented, but in reality there are challenges in building AI at scale. It is pertinent to note that building AI at scale is crucial to make it available for the mass.
Besides, he said that India has also witnessed a balance move from global tech companies to homegrown tech startups backed by venture capital firms. He hailed startups like Meesho, PhonePe, PhysicsWallah, Zepto, Rapido and Urban Company for making strides in their respective sectors.
A month ago, Nilekani said that India will be home to more than 10 Lakh startups by 2035.
He added that while the Indian economy grows at 8%, the number of startups in the country would grow by a 20% compounded annual growth rate (CAGR) and reach the 10 Lakh mark in the next decade.
He further said that initial public offerings (IPOs) would also fuel the rise of the Indian startup ecosystem.
The enterprise software landscape is being transformed by adoption of AI and the rise of new use cases driven by changing consumer behavior. As a result, a slew of startups integrating generative AI and new-age tech advancements in their offerings have gained traction from investors.
Earlier this year, enterprise AI startup Atomicwork raised $25 Mn in a Series A funding round co-led by Khosla Ventures and its existing investor Z47 Partners.
After that, another such startups Singulr AI bagged $10 Mn in a seed funding round co-led by Nexus Venture Partners and Dell Technologies Capital.
It is pertinent to note that Generative AI (GenAI) in India is on track for an exponential growth with projections estimating a market value of over $17 Bn by 2030.
The GenAI startups have collectively raised more than $1.2 Bn in total funding since 2020, as per Inc42’s report.
