The Indian government is stepping up its efforts to support domestic AI development, with plans to back homegrown large language models (LLMs) from Indian startups and companies. This announcement comes after Chinese company DeepSeek's low-cost AI model disrupted the tech industry, causing a massive market dip for American tech giants.

 

Union IT Minister Ashwini Vaishnaw shared that six Indian companies are set to release their AI foundation models within the next ten months. He emphasized the importance of algorithmic efficiency, noting that Indian researchers have been working on similar concepts, which will enable India to develop world-class models at a fraction of the cost and time seen globally.

 

Foundation models are the core of advanced AI systems like ChatGPT. They are versatile algorithms that learn from vast amounts of data and can be fine-tuned for a variety of applications, from coding to language understanding.

 

This initiative follows DeepSeek’s release of its open-source R1 model, built at a fraction of the cost of U.S. models like GPT-4. Vaishnaw added that DeepSeek’s models would be hosted on Indian servers to ensure data security and privacy, although he didn’t confirm if discussions with the company were ongoing.

 

To foster local AI innovation, IndiaAI is rolling out a process to accept proposals from startups, which will be evaluated monthly. Successful applicants will receive funding or investment opportunities. The government is particularly focused on building models that reflect India’s unique languages, culture, and societal challenges.

 

In support of the AI push, the Ministry of Electronics and Information Technology (MeitY) has also announced a partnership with ten companies that will provide computing power across various GPU categories. These companies have pledged over 18,000 GPUs, surpassing the target of 10,000, to help accelerate AI development under the IndiaAI Mission.