India to Develop Native Generative AI Models in Six to Eight Months, Says Ashwini Vaishnaw

India will build native generative artificial intelligence (AI) models within six to eight months, the Union Minister of Electronics and Information Technology (IT) Ashwini Vaishnaw announced on Thursday. The initiative was announced at the Utkarsh Odisha Conclave, where the minister also revealed that the DeepSeek-R1 AI model has been implemented on local Indian servers and that experts are currently evaluating the model's technical report. Notably, the announcements came just a week after the conclusion of the 55th annual meeting of the World Economic Forum in Davos-Klosters, Switzerland where India was also a participant.

India Gearing Up to Build Native AI Models

According to a Moneycontrol report, Vaishnav made several AI-focused announcements at the Conclave while speaking to the press. The IT minister stated that India has secured more than 18,600 high-end GPUs to build the AI infrastructure and scale compute. This will become the backbone of the generative AI models that will be developed this year.

"We believe that there are at least six major developers who can develop AI models in six to eight months on the outer limit, and four to six months on a more optimistic estimate,” Vaishnaw was quoted as saying.

The Union minister also highlighted that there is a critical need to develop a common compute facility as it is the most important pillar to build a strong AI ecosystem. This initiative will be part of the India AI Mission and will support the computational requirements of researchers, startups, and academic institutions.

Detailing the planned compute facility (cloud-based servers that are geared to run and handle AI inference), Vaishnaw stated that it has exceeded initial expectations with the country securing more than 18,600 GPUs. Among these are 12,896 Nvidia H100 GPUs, 1,480 Nvidia H200 GPUs, and 742 AMD MI325 and MI325X GPUs. Originally, the target was to procure 10,000 high-end AI chipsets, the IT minister said.

“DeepSeek AI was trained on 2,000 GPUs, ChatGPT was trained on 25,000 GPUs, and we now have 18,000 high-end GPUs available. India now has a robust compute facility that will support our AI ambitions,” he added. Notably, once deployed, the Indian AI models will be accessible to everyone. The minister didn't specify if the models will be open-source, or if they will be available via platforms and application programming interfaces (APIs).

On DeepSeek, Vaishnaw stated that India has already implemented the DeepSeek-R1 AI model and experts are currently evaluating the technical report to gain a better understanding of it, as per a CNN-News18 video footage.

Focusing on AI safety and security, Vaishnaw said that India will establish an AI safety institution with a hub-and-spoke model. Unlike countries where a single body oversees AI regulation, India will allow multiple institutions to collaborate and build safety tools and frameworks.