India is at the limit of a major AI breakthrough with the development of its first LLM, like ChatGPT and DeepSeek. According to Union Minister of Electronics and Information Technology Ashwini Vaishnaw, AI model development is already on track and will be ready in 10 months, completing the first vital step in India for it to be self-sufficient in AI.

What's behind India's AI push?

Speaking at the Utkarsh Odisha Conclave, Vaishnaw said that thorough groundwork has already been done. The framework of India's AI model is there, and it is only efforts in developing the system based on the needs for the country that remain to be seen.

"Over the last 1.5 years, our teams have been working closely with startups, researchers, and professors. Today, we are calling for proposals to develop our own foundational model. The model will take care of the Indian context, languages, and culture, and will be devoid of biases," Vaishnaw said.

India's AI infrastructure: More GPUs, more power

Advanced computing infrastructure is one of the most crucial factors advancing India's AI ambitions. It has already achieved more than double its original aim of securing 10,000 GPUs, as it currently boasts 18,600 GPUs. These powerful computing units will form the foundation on which India trains its AI model.

DeepSeek AI was compared to being trained on 2,500 GPUs. ChatGPT used approximately 25,000 GPUs. India has more than 15,000 high-end GPUs, such as NVIDIA H100 and H200, along with MI325 models. This puts India in a good position in the race with other countries to produce a language model of international standard.

Common computing facility for startups and researchers

In addition, to democratize the development of AI, the government launched a common computing facility for AI startups, developers, and researchers. It has 18,000 GPUs with the potential to cut off the financial barrier that hinders the participation of smaller players in such cutting-edge computing resources. Presently, 10,000 GPUs are already activated while others will soon be activated.

This initiative would allow Indian innovators to participate in AI progress without the prohibitive costs that generally come with such high-end infrastructure, Vaishnaw highlighted.

AI model expected in 4 to 10 months

India is working on an AI model. It has six big developers on it. The government expects the first versions of this model within 4 to 10 months. The model will be made such that it would be suitable to India's linguistic and cultural diversities and therefore efficient as well as inclusive.

Vaishnaw expressed confidence in the project, stating that improvements in algorithmic efficiency will allow India to develop a competitive AI model within a shorter time frame. With strong government backing, state-of-the-art infrastructure, and a clear focus on inclusivity, India is set to make a mark in the global AI landscape.