Tandon, who advises Fortune 500 companies worldwide on implementing AI, said in an exclusive interview: “We cannot expect the big IT firms in India, in their current shape and form, to be at the forefront of this AI change.”
He points to the US as an example. “Look at the US. Microsoft did not drive this change, nor did Oracle or SAP. The change came from OpenAI and Anthropic. So the new wave of change is most likely going to be fuelled by people who are not bridled by the past,” he said.
Tandon argues that these new companies will focus on innovation, unlike traditional IT services firms, which have largely competed on their ability to deliver services at lower costs. “India has a huge young population and an open environment, and there is scope for new players to come in and grab this opportunity,” he added.
On the emerging harmony among major US AI companies over safeguards for AI, despite their recent disagreements over the contentious issue, Tandon said the push for guardrails was driven partly by concerns about competition from open-weight models.
“The reason is they do not want to get overtaken by the open-weight models. They want to impose policies so that users cannot use open models without a huge amount of paraphernalia around them and guardrails. They have put in billions of dollars of investment over the past three years, but now, with technology advancing so much and open models coming at a fraction of the cost, they are asking: how do I recover the cost or make money?” he said.
Tandon added that excessive guardrails could push a large part of the world towards open models.
He pointed to China, France and a growing number of other countries that are betting on open models. Several US companies, too, are developing and backing such models.
Can guardrails designed today remain effective given the pace of AI development? Tandon said the speed at which new models are emerging makes it impossible to predict every risk and put safeguards in place at the outset.
“AI works at a speed that we have not seen before. So while speed can give us a huge amount of benefit, if things go wrong, they will go wrong very fast,” he said.
Tandon argued that every country must adopt AI at a pace it can manage. “The pace might be different in India, Europe or China. Also, the worrying factor is that AI can be used for good, but also for bad — and those guys are not waiting for policy to come,” he said, referring to those who might misuse the technology.
Tandon expects enterprises to adopt hybrid AI models, combining closed and open models for different applications.
“The right AI model to use could mean multiple things. I could have three models which can do the same thing. One will give me an answer in three seconds. Another will take 30 seconds and cost one-third the price. So unless it’s a life-changing decision that I need to make in a few seconds, I would go for the cost-effective model,” he said.
India’s IT giants risk being left in AI’s rear-view mirror
- India’s IT services majors may struggle to lead AI shift in their current form
- US AI firms’ push for safeguards comes amid concerns that cheaper open models could undercut their investments and business models
- China, France and other countries are backing open models, with several US companies also developing them
