India should not chase the costly LLM race; instead, he should focus on practical models of artificial intelligence: Mohandas Pai
Aarin Capital TV chairman Mohandas Pai and ToneTag founder and CEO Kumar Abhishek launched eKosha, ToneTag’s first voice-based AI merchant assistant that transforms payment devices into always-on banking touchpoints, in Bengaluru on Thursday. | Photo credit: ANI
India’s current constraints on capital and computing infrastructure make hands-on adoption of AI a more sensible priority before chasing expensive frontier large language model (LLM) races, advised Mohandas Pai, chairman of venture capital firm Aarin Capital and former CFO of Infosys.
He was speaking at the launch of fintech firm ToneTag’s first voice merchant banking platform eKosha here on Wednesday.
Mr. Pai further said that India should focus on deploying AI to solve real-world problems along with improving productivity. “India’s AI success would ultimately not be measured by the number of frontier models it creates, but by how effectively AI improves the lives and productivity of ordinary citizens, entrepreneurs and small businesses,” he said.
He said public discourse lacked recognition of the enormous investment and infrastructural effort required to build LLMs, and the country was overly focused on whether it was building LLMs.
“We see so much criticism in this country of people suddenly becoming experts in AI and technology, saying why don’t you build an LLM or why don’t you do this? It costs $35-40 billion to build an LLM and nobody in this country is writing that kind of check,” Mr. Pai said.
He also said that access to massive hyperscale computing infrastructure is essential to developing frontier models of artificial intelligence. To create artificial intelligence, the country needed a hyper-cloud and it has a capacity of two gigawatts, while the US has 40 GW and the US would invest about 3 trillion dollars in the next two years. “We are just starting our voyage in India,” he noted.
Voice AI will be the future
Voice would be the next major interface for digital services, replacing many app-based interactions by allowing people to interact with technology in the most natural way, opines Mr Pai, adding: “Voice is the primary means by which human beings communicate.”
The practicality of voice interfaces is growing exponentially across Indian languages due to advances in speech recognition, speech-to-speech and small language models, he noted.
Artificial intelligence and robotics are likely to automate a significant portion of human activities in the coming years and it was imperative to equip India’s 60 million-plus SMEs with affordable productivity-enhancing technologies, he advised.
Published – 02 Jul 2026 23:10 IST