3 min readSep 18, 2025 08:55 PM IST
First published on: Sep 18, 2025 at 08:55 PM IST
In line with its aim to support the development of domestic artificial intelligence (AI) models, the government has selected eight more startups to build indigenous Large Language Models (LLMs) and sector-focused Small Language Models (SLMs), which includes the likes of Tech Mahindra, Fractal Analytics, Avatar AI, Zeinteiq Aitech Innovation, and the Indian Institute of Technology-Bombay consortium’s BharatGen, among others.
Earlier, the government had selected four startups for building such models, including Sarvam, Soket AI, Gan AI, and Gnani AI. Aside from those mentioned above, the government has also selected Genloop Intelligence, NeuroDX, and Shodh AI to build LLMs as well as SLMs.
“Our selection of partners for developing models is also changing…that’s adapting to the latest thought in the technical world that models which are focused on solving the problems of a particular sector will be more effective, have more utility, and be more in demand,” Union Electronics and Information Technology Minister Ashwini Vaishnaw told reporters.
The minister said that while it will continue to focus on large models, it will be in sector-specific smaller models where India could have an edge, especially in sectors like education, healthcare, agriculture, materials science, manufacturing, and law, among others.
Vaishnaw said that the government has in the past focused on building technologies such as UPI, that can be shared with other countries, and it plans to take a similar approach with its investments in AI. To that end, he said that the government was looking at ways to share the graphics processing unit (GPU) compute capacity that has been created so far under the IndiaAI Mission with some countries of the global south. So far, more than 38,000 GPUs have been onboarded under the flagship mission to augment the country’s supercomputing capacity.
Vaishnaw was speaking at an event announcing the upcoming IndiaAI Impact Summit, the fourth iteration of the global AI symposium, which has so far been held in the United Kingdom, South Korea and France. In February 2026, India will host the fourth leg of the summit, and Vaishnaw said the country will have a particular focus on AI safety, a subject which seemed to be largely absent in the summit that took place in France.
Ahead of the summit, India also plans to launch an AI governance framework, a guiding document for tech companies to adhere to while launching technological solutions in the country. Through the framework, the government will also hope to build a global consensus on certain common minimum principles that could be applicable to companies around the world.
“As we interact with global leaders, one thing that is coming out in all dimensions is that the world has to come together for a common governance framework on AI. It is too important a technology, and the impact is so profound that we must all come together and look at a common governance framework,” Vaishnaw said.
The government will also establish more than 500 AI data labs across the country, particularly in tier-II and tier-III towns and cities, for skill development in advanced technologies such as AI, machine learning, and others, he said, adding that a total of Rs 500 crore is likely to be spent on the project.
Earlier, the government had selected four startups for building such models, including Sarvam, Soket AI, Gan AI, and Gnani AI. 
Share
