Govt to Extend IndiaAI Incentives to 8 Homegrown Foundational-Model Teams

The government has shortlisted close to eight domestic firms developing foundational AI models for an extension of incentives under the IndiaAI Mission, Economic Times reported. Union minister Ashwini Vaishnaw is expected to formally announce the second-round beneficiaries next week.

Among the likely recipients is BharatGen, a government-backed consortium anchored by IIT Bombay, which recently released a bilingual foundation model called Param-1 (about 2.9 billion parameters and trained on large English–Hindi corpora). Sources say BharatGen’s entry would mark a high-profile addition to the roster of supported projects.

IndiaAI Funding Framework

The selection comes under a broader incentive window worth ₹1,500 crore announced by MeitY to support organisations building foundational models from the ground up, via grants, compute credits and other support, as part of a drive to create sovereign AI capabilities. The call for proposals attracted hundreds of applications earlier this year.

This is the second phase of approvals after an initial cohort: earlier in the mission Sarvam, Soket Labs, Gnani.ai and Gan.ai were approved to receive subsidised GPU compute and other assistance to develop indigenous models. The new round will broaden that group if finalised as expected.

GPU Capacity

IndiaAI’s compute programme has an ambitious target, about 34,333 GPUs empanelled under the mission, but officials and industry sources warn that only about 17,374 GPUs are currently installed and operational, creating a short-term “supply crunch” as more teams seek subsidised access. That gap has prompted internal debate inside MeitY about prioritising inferencing-efficient hardware as well as training capacity in future procurements.

The mission has empanelled cloud and GPU-as-a-service providers to run a common compute facility intended to make GPU hours available at subsidised rates, a central plank of the ₹1,500 crore support plan. Officials say the facility is designed to democratise access for startups, academia and researchers while keeping costs low.

Backing homegrown foundational models addresses strategic goals, data sovereignty, language coverage and locally relevant capabilities, while reducing reliance on foreign models and infrastructure. But the effectiveness of the policy will hinge on whether compute supply scales fast enough, and whether selected teams can translate subsidised GPU access into performant, safe models that serve Indian use cases.