⚠️ DISCLAIMER: Ministry of Cyber Affairs is an independent news, guides and directory service on cybercrime and cybersecurity, for victims and investigators. It is NOT an official government or ministry portal.
Website Scam CheckerNew
Back to News

India to build Sovereign AI Capabilities for Army, Navy and Air Force in collaboration with Sarvam AI

27 October 2026
Source:HQ Integrated Defence Staff (X, 26 Oct 2026); Sarvam statement, 27 Oct 2026; PIB
Share
India to build Sovereign AI Capabilities for Army, Navy and Air Force in collaboration with Sarvam AI

Headquarters Integrated Defence Staff and Sarvam AI have signed a Memorandum of Understanding to collaborate on AI applications, research, knowledge management, training and decision support across the Indian Army, Navy and Air Force. The partnership will stress secure, sovereign and scalable AI, data sovereignty and responsible AI. Sources say it runs for five years under Sarvam's defence division, Chanakya.

India's Integrated Defence Staff and homegrown AI company Sarvam have signed a Memorandum of Understanding (MoU) to build artificial intelligence capabilities for the Indian Army, Navy and Air Force, with data sovereignty and responsible AI as stated priorities.

What Was Announced

HQ Integrated Defence Staff (HQ IDS) said on 26 October 2026 that it and Sarvam AI signed the MoU to collaborate on AI applications, research, knowledge management, training and decision support across the three services.

It called the partnership "a significant step towards strengthening sovereign AI capabilities for India's Defence Forces" and said it will focus on "secure, sovereign and scalable" AI, with strong emphasis on data sovereignty and responsible AI-based solutions.

Photographs released with the announcement show the document exchange between an Indian Air Force officer in uniform and a Sarvam representative, followed by a group photograph of senior tri-service officers and company staff.

Neither signatory is named in the post.

Sarvam's Account

Sources say Sarvam described the MoU on 27 October as an umbrella framework for AI projects across the three services and the joint, tri-service and integrated organisations under HQ IDS.

Sources say Sarvam will design, develop and deploy AI applications for operational, training and decision-support needs, and provide follow-on technical support.

Sources say the services will share the same foundation models, agentic infrastructure, and geospatial and perception systems, so lessons from one service's projects can feed the next.

Focus areas reportedly include decision support, target acquisition and data fusion across the sensor-to-shooter loop, with human oversight retained over operational decisions.

Sources say the work will be carried out by Chanakya, Sarvam's firewalled national security and defence division, using models such as Sarvam 105B.

Sarvam co-founder Pratyush Kumar is quoted as saying: "Sarvam exists to build AI in India, for India."

Sources say the MoU runs for five years.

Sources say the financial value and individual projects were not disclosed.

Background

Sarvam is a Bengaluru-based company and, per the Press Information Bureau, one of 12 organisations selected under the IndiaAI Mission to develop indigenous foundational models, with support of ₹246.72 crore.

The Sarvam 30B and 105B models were trained in India on IndiaAI Mission compute and released as open weights in March 2026; Sarvam 105B is available on Hugging Face.

The MoU builds on earlier work.

On 18 August, the Defence Services Staff College, Wellington launched DRONA 2.0, a multimodal AI platform integrating Sarvam 105B, and signed an MoU with Sarvam.

HQ IDS and the Defence Cyber Agency audited and certified the platform before deployment.

The Ministry of Defence already has a structure for military AI.

A 2019 executive order created the Defence AI Council and the Defence AI Project Agency, and an AI roadmap was finalised for each defence public sector undertaking, identifying 70 AI projects.

Sarvam's Upcoming Flagship Model

Sarvam has said it is building a trillion-plus-parameter model from scratch in India, announced at its Epoch developer conference on 30 July 2026 and aimed at coding, cybersecurity, science and simulation.

It has no public name yet.

Sarvam has said it will go live within six months, and sources say co-founder Pratyush Kumar has since pointed to February 2027.

Sarvam has said it holds 2,000 Nvidia Blackwell GPUs and needs about 10,000 for training.

Whether IndiaAI Mission compute will be used is unconfirmed, though the 30B and 105B models were trained on it.

Language coverage and any defence role for the new model have not been published; Chanakya currently uses Sarvam 105B.

More in AI Updates

Related articles

AI Updates

GPT-Red: OpenAI's Internal Attack Models for Automated Red-Teaming

OpenAI’s public catastrophic-risk report for June 4–September 4, 2026 describes GPT-Red as restricted internal models that generate attacks—especially against prompt-injection defenses—to test and train stronger safeguards. They sit with other restricted checkpoints such as cybersecurity models and GPT-Rosalind, typically in higher-isolation L3 environments.

9 Nov 2026
AI Updates

How does Google's Gemini 4 train its model ? Trillion-Token Scale, Reasoning and Synthetic Data

Google's EU Public Summary of Training Content (updated 13 October 2026) says the Gemini 4 family was trained on more than 10 trillion text tokens, over 1 billion images and over 1 million hours each of audio and video. It reveals reinforcement learning for reasoning and theorem-proving in Gemini 4 Argon, synthetic data from Gemini 3, Gemini Apps user chats, and robots.txt and Google-Extended controls, with no country breakdown.

9 Nov 2026
AI Updates

Maharashtra Puts Sarvam's Sovereign AI on 2,500 Officials' Desks as Indian Governments Turn to Home-Grown AI

About 2,500 Maharashtra government officials can now use Sarvam AI's Indus workspace, launched as MahaINDUS by Chief Minister Devendra Fadnavis, to draft documents, summarise files, search Government Resolutions and evaluate tenders in 22 languages. The state also signed an Enterprise Agreement with Sarvam's parent Axonwise. Officials say data stays in India and is not used to train AI models.

29 Oct 2026