Under its computer program, the government is deploying more than 18,000 GPUs, which includes nearly 13,000 high -end H100 chips, including an selected group of Indian Startups, which currently includes Sarwarram, Aparwal’s Sort Labs, GNANI AIAnd Song Ai.
The mission also includes plans to launch a national multi -linguistic data set repository, set up AI lab in small cities, and fund Deep Tech R&D. The broader purpose is to equip the Indian developers with the infrastructure needed to build a global competitive AI and to ensure that the results are based on linguistic and cultural facts of India and the global South.
According to Abhishek Singh, an Indian CEO and a Meeting officer, it is expected that the widespread pressure in India’s deep tech will increase about $ 12 billion in research and development investment over the next five years.
This includes about $ 162 million by the Indian mission, allocating about $ 32 million for direct startup funds. The National Quantum Mission is supporting $ 730 million to support India’s ambitions in Quantum Research. In addition, the national budget document for 2025-26 announced a $ 1.2 billion deep tech fund of funds to uphold the initial phase innovation in the private sector.
It is expected that the rest, about $ 9.9 billion, will come from private and international sources, including global technology leaders such as corporate R&D, venture capital firms, high networking people, donors, and Microsoft.
Indian has now received more than 500 applications from startups suggesting issues in areas such as health, governance and agriculture.
“We have already announced support for Saram, and 10 to 12 more startups will be fully funded for basic models,” says Singh. Selection standards include training data access, talent depth, sector fit, and scalebuability.
Open or closed?
However, the Indian program is not without conflict. Sarwarram is being built as a closed model, not open source despite public tech roots. It has given rise to a debate about the proper balance between private enterprises and the good of the public.
“Real sovereignty should be rooted out of openness and transparency,” says Mohani, an AI policy expert. He pointed to the DEP Sak-R1, which, despite its 236 billion parameter size, was made independently for commercial use despite the size.