AI Impact Summit 2026: Ex-HCL Tech CEO Weighs In On The Great Indian Data Debate

At the AI Impact Summit 2026, Ex-CEO HCL Tech Vineet Nayar noted that India needs to maintain a cautious approach when it comes to the ownership of data.

While countries like China, Italy and Netherlands limit how AI firms access personal data for AI training, India with a population of 1.4 billion people and the large-scale implementation of AI models could become a major provider of high-quality datasets to improve AI systems.

However, the debate begins from should one allow global LLM products to train on Indian data sets.

"The LLM models which exist worldwide are far superior than the Indian models. Unfortunately, in India, and I am to blame for that, and Indian IT is to blame, we never develop products. So therefore, we do not have SLMs and LLMs, which are world-class. So on one side, we have global LLM products, which they are coming to India and training on our Indian data," he said.

"But on the other side, if we don't allow that, then we have the data, but we don't have the LLM models. So how do we encourage technology companies to develop the LLM models? And therefore, this needs radical strategic thinking and a very important aspect," Nayar noted at one of the largest AI summit being held in the Global South.

Also Read: Bloodbath For IT Stocks? - Nifty IT Declines 32% From Its Record Peak

The Microsoft Lesson For India's AI Path Ahead

"I must remind you, I come from a time when the computer industry came in and Microsoft DOS operating system was thrown free in this country, and HCL had an operating system, Infosys had an operating system, TCS had an operating system, all of us had an operating system," Nayar said.

When Microsoft entered India, "the DOS was pirated, and everybody had a PC with the pirated DOS". This resulted in the three companies and their operating systems getting “killed”. 

"DOS became the operating system, and then they started charging billions and billions of dollars, and the country lost the capability of developing operating system," he said.

Meanwhile, access to Indian data remains critical for the success of any AI models, which has resulted in free use access to AI models in the South Asian nation.

Currently, the most successful LLMs have been trained on nearly 15 trillion tokens or datasets, signalling the value of date to develop AI models.