The Centre is likely to push for wider use of indigenous AI models, large language models and AI agents in critical government sectors such as defence and aerospace, senior government officials told Business Standard on 9 November 2026. The move could see some central and state departments asked to avoid frontier AI models and agents built by non-Indian companies altogether.

The plan marks a shift for sovereign AI in India, from building homegrown models to actually requiring them where national security and sensitive data are involved.

What the government is considering

• Preference for Indian models: critical sectors, starting with defence and aerospace, may be asked to prioritise AI systems developed under the IndiaAI Mission.
• Limits on foreign frontier AI: some departments could be told to stop using frontier models and agents from non-Indian companies completely.
• Review of deployments: sectors such as defence, already building their own air-gapped LLMs that run on the edge, have been asked to study how IndiaAI models and agents can be deployed.
• Wider model scope: officials say they are in constant talks with the companies building models to expand what those models can do.

“The idea is to have enough technological capability to use these models in crucial sectors,” an official said, adding that some domestic models have reached up to 80% of the capability of leading frontier AI models.

The IndiaAI model pipeline behind the push

Under the ₹10,372 crore IndiaAI Mission, the government has backed 20 foundation models, a mix of general-purpose LLMs and sector-focused small language models (SLMs), built by 12 startups. Sarvam and BharatGen have already released several models, and the mission recently released Soket’s EKA-1 and ZenteiQ’s Dhir 2B and Dhir 5B models, which MeitY says can run in air-gapped environments. A video-generation model called Varya has also been released.

Gnani, Gan AI, Fractal, Tech Mahindra and Shodh AI are expected to release their LLMs and SLMs over the next 12 to 15 months. Government-backed models are also being listed on AIKosh, as with CoRover’s three BharatGPT models.

Defence is already moving towards sovereign AI

The armed forces have been among the first to act. Sarvam AI and HQ Integrated Defence Staff signed an MoU to build sovereign AI for India’s armed forces, and BharatGen has added former military chiefs to the advisory board guiding its move to deployment. In October, Defence Secretary Rajesh Kumar Singh said India was developing its own military AI layer for decision support, target acquisition and data fusion, according to Moneycontrol.

Hardware remains the weak link. Indian models still run on foreign GPUs, which is why MeitY is targeting an indigenous GPU chip in about 24 months.

What it means for CIOs and technology vendors

If the policy is formalised, technology leaders in defence, aerospace and other critical public bodies will need to map where foreign frontier models are used today, plan migrations to Indian models, and test whether a model with about 80% of frontier capability is good enough for each task. For Indian AI startups, it opens a large anchor market; for global AI companies, it could narrow access to the most sensitive government work. The full report is on Business Standard.

Key takeaways

• The government is likely to push indigenous AI models in critical sectors such as defence and aerospace.
• Some departments may be asked to avoid frontier AI models and agents from non-Indian companies.
• The ₹10,372 crore IndiaAI Mission backs 20 foundation models from 12 startups.
• Officials say some Indian models reach up to 80% of leading frontier model capability.

Be a part of Elets Collaborative Initiatives. Join Us for Upcoming Events and explore business opportunities. Like us on Facebook , connect with us on LinkedIn and follow us on Twitter.

"Exciting news! Elets technomedia is now on WhatsApp Channels Subscribe today by clicking the link and stay updated with the latest insights!" Click here!