Big AI models will be commoditized — the value will come from the applications, Infosys chair says

Massive artificial intelligence models built by the likes of OpenAI will eventually become commoditized with the next wave of value in the technology coming from the applications that are built on top of the models, the co-founder and chair of Infosys told CNBC on Tuesday.

Large language models are the AI models trained on amounts of data and they underpin applications like ChatGPT. Companies such as OpenAI and Meta are among the biggest players in this space.

Nandan Nilekani, co-founder of Infosys, said there are going to be many different LLMs around the world.

“There are many companies now coming up which are building India-specific LLM solutions for Indian languages … so I think what’s going to happen is that because AI is finally dependent on the data it is  trained on, every part of the world which has unique data will be required to do something about training the models for that data,” Nilekani told CNBC’s “Squaw Box Europe” on Tuesday.

“Ultimately, the models will become more commoditized and the value will switch to the application layer and the whole stack.”

Applications can be built on top of LLMs, and according to Nilekani, the next wave of value from AI will derive from those applications and other related technologies.

Infosys is one of India’s biggest technology companies offering outsourcing and consultancy services.

Discussing trends in AI, Nilekani said that the integration of AI into businesses, also known as enterprise AI, “is a longer cycle than consumer AI.”

“Consumer AI you can get up a chatbot and start working. Enterprise AI requires firms to reinvent themselves internally. So it’s a longer haul, but definitely it’s a huge thing happening right now,” Nilekani said.