India to host Chinese AI model DeepSeek on local servers, plans its own foundational model

The minister also revealed that India is set to develop its own foundational AI model in the coming months, underscoring the country's ambitions in the field.

Additionally, he announced the empanelment of 18,693 graphics processing units (GPUs) under a shared computing facility.

Vaishnaw made a series of announcements aimed at positioning India as a key player in the global AI landscape, including plans to establish an AI safety institution.

Vaishnaw stated that the Prime Minister's economic vision focused on making modern technology accessible to everyone.

He also claimed that India currently had the most affordable computing facility.

In March, the government committed over Rs 10,300 crore to the IndiaAI mission, a major investment initiative that includes funding for AI startups and the development of domestic AI infrastructure.

Addressing concerns about the scale of investment, Vaishnaw cited DeepSeek’s rapid progress, noting that some people questioned the government's financial commitment to the IndiaAI mission.

He pointed out that DeepSeek had achieved significant results with an investment of $5.5 million, producing a highly powerful model through effective utilisation of intellectual resources.

His remarks, made at an event in Odisha on Tuesday, appeared to reference OpenAI CEO Sam Altman’s skepticism last year about India’s ability to develop a major AI model.

DeepSeek, developed by a Chinese startup and released recently, has gained significant attention in the AI industry.

The model, known as DeepSeek R1, has been praised for its efficiency and effectiveness, particularly in complex reasoning tasks.

Its app downloads recently surpassed OpenAI’s ChatGPT on Apple’s App Store.