India Needs To Build Own LLMs To Reduce Reliance On Other Countries: HCLTech CEO

IT giant HCLTech CEO C Vijayakumar said that India needs to build its own large language models (LLMs) to limit reliance on open-source AI models and have a long-term competitive advantage. 

Speaking at Nasscom Technology and Leadership Forum 2025, Vijayakumar said that building models on top of existing models may work for now, but there is a possibility that these open-source models may not remain accessible in the longer run. 

“We can make models on top of the existing global models, but we shouldn’t assume that the open-source models will continue to be open source,” he said. “These are going to be the coins on which the geopolitics will be played on.”

He further warned that the leading AI companies like OpenAI , Meta, or DeepSeek may limit the usage of their AI models at some point in the future. To have a competitive advantage in the long term, India needs to find ways to create an economic training infrastructure to train the models. 

Chiming in, IT major Infosys chief Salil Parekh said, “The cost of building models is falling, so when the time is right, we should make sure that we make the right investments and build small language models.” 

Parekh said that India needs to ensure it is the leading light of generative AI by making investments to create foundational models. 

The development comes amid a growing debate about the need for India to have its own LLM. While Indian IT giants like TCS, Infosys, HCLTech and Wipro are banking on GenAI to grow their businesses, they have refrained from making their own LLMs so far.

Last month, Perplexity founder and CEO Aravind Srinivas said that India should build its own foundational models. This came a few months after Infosys cofounder Nandan Nilekani said that India shouldn’t focus on building foundational models from scratch.

Meanwhile, amid the rising popularity of China’s DeepSeek, the IndiaAI Mission, last month, floated a proposal inviting applications from startups, researchers, and entrepreneurs to collaborate on building foundational AI models trained on Indian datasets.

The Centre has received about 67 proposals for building indigeneous foundational models.

At the heart of this push is the growing adoption of GenAI in the country. This has also spawned the rise of more than 200 GenAI startups that have raised more than $1.2 Bn since 2020.