India's Yotta Data Services, which is developing a $2 billion artificial intelligence hub utilizing Nvidia's chips, reported that demand for graphics processing units (GPUs) in the country is outpacing supply as domestic AI models prepare to scale and the local user base grows rapidly. Currently, India lags behind the U. S. and China in creating a native AI foundational model and lacks extensive domestic AI infrastructure. However, this situation is beginning to change. Last week at the India AI Summit, several Indian companies introduced early or limited versions of their AI models, including Sarvam AI’s Indus chatbot. “We’re gradually rolling out Indus with limited computing capacity, so you might face a waitlist initially. We plan to expand access over time, ” said Pratyush Kumar, co-founder of Sarvam AI, in a post on X. The majority of Indian AI models debuted at the summit were trained on Nvidia GPUs hosted at Yotta’s facilities, according to Sunil Gupta, co-founder, managing director, and CEO of the Mumbai-based data center company, during an interview on CNBC’s Inside India on Thursday. Yotta, which started sourcing Nvidia GPUs in 2023, now controls about 60% to 70% of India’s GPU capacity, Gupta stated. He also mentioned that demand is expected to rise from global AI companies as their user base in India expands. Recently, major U. S.

tech companies such as OpenAI, Google, and Perplexity have made their AI tools available to millions of Indian users at low or no cost. Among hyperscalers, Google has confirmed plans to invest $15 billion in building a data center hub in southern India, while Microsoft will invest $17. 5 billion to expand its data center presence. Last week, OpenAI became the first client of Tata Consultancy Services’ data center business in India, signing a contract for 100 MW of capacity, with an option to scale up to 1 GW. “Through OpenAI for India, we are partnering to build the infrastructure, skills, and local collaborations needed to develop AI with India, for India, and in India, ” said Sam Altman, OpenAI’s CEO, in a statement on February 19. As the Indian user base of major global AI companies grows, Gupta noted they will increasingly require local data centers and GPU capacity. Yotta plans to finance further GPU acquisitions through a $1. 2 billion to $1. 5 billion pre-IPO funding round and aims to go public within the next 12 months, he added. India’s total data center capacity is expected to reach 1. 93 gigawatts in 2025, nearly doubling to 4 gigawatts by 2028, according to a February 20 report from Nomura. During the AI Summit, many firms announced intentions to invest $277 billion over the next five to seven years, with the majority of funds directed toward building AI infrastructure in India, the brokerage noted. “Most of these investments will go into data centers, with both domestic and U. S. companies leading hyperscale developments, positioning India as a key U. S. technology partner, ” the brokerage stated.