Sikka, who was among the early backers of OpenAI when it was still a non-profit, warned that the country cannot afford to be a passive user of AI tools built elsewhere. “We are too big. We are too important. And the stakes are too high for us to stay out of any one of those layers,” Sikka said, adding that AI is already changing how software is written, how businesses operate, and how productivity is measured.
In his view, India should not just consume AI but create it, scale it and export it. That includes building its own core models, tools on top of them, and training millions of people to use AI in meaningful ways. “We have to play and we have to win,” he added.
He gave an example of how a system that once took 15 people and nine months to build was recently recreated by one person in just two weeks using AI. That, he said, shows how quickly the ground is shifting. "It's not a matter of when it's coming. It is here already."
So, how quickly we can transform ourselves towards becoming providers of AI from people who are disrupted by AI is the most important equation for the time ahead.
Sikka said India’s readiness challenge is twofold. “It is twofold. One is the skill matter… The other part of it… is that it has a disruptive effect,” he said.
Skills, he said, are a major bottleneck. Only a tiny fraction of the global population can actually build advanced AI systems today. The number of people who can build foundation models, he said, is “less than 25,000 on the whole planet.” But these are not mystical skills and can be taught, learned and scaled. India just has to move fast to make that happen.
The second challenge, he said, is how deeply AI can disrupt existing jobs and business models — especially in software services. “Large language models are unreasonably effective at writing software,” Sikka said, warning that it will directly affect people who write and maintain software for a living. India’s vast services industry will need to “transform very quickly” to adapt to this shift
Sikka has seen these shifts coming for years. In 2015, he was among the earliest backers of OpenAI, when it was still a non-profit. “We did give a donation to OpenAI… and we were one of the earliest backers,” he said, adding that he also served as an advisor. At the time, he said, it was already clear where AI was headed. “You could see that impact of AI on the work that we used to do was going to be profound,” he said.
But while enthusiasm around AI is high today, Sikka is sceptical about current valuations. He said the numbers don’t quite add up. Pointing to Nvidia, he said the company is generating about $250 billion a year, and when you add CPUs, memory and networks, the total AI infrastructure spend reaches around $750 billion annually. “This is real,” he said.
But he questioned where the revenue to justify this spending will come from. “That 800 billion means that there should be 200 billion in revenue to pay for all this. And where is the 200 billion?” he asked, adding, “The current reality economics don’t add up to what has been spent.”
The price-to-earnings ratio for the S&P 500, the index of America's biggest stocks, would fall from 22.7 to 21.9 if the likes of Nvidia, Palantir, Tesla, and other new-age tech stocks are taken out of the calculation, a recent analysis by Bloomberg's Jonathan Levin showed.
The average valuation multiple for the last 20 years for the American index was 16%. While many like Gary Marcus, an Cognitive scientist-turned-AI researcher, consider it an AI bubble, there are others who beg to differ.
Despite the risks, Sikka remains optimistic. He believes India still has time — but not much of it. Using a cricket analogy, he said, “If it was a T20 match, we are in the first over.” That means the game is still wide open, but action has to start now.
For full interview, watch accompanying video
