Sarvam AI is at the centre of India’s effort to develop artificial intelligence systems for sensitive government and military applications. A reported five-year defence AI agreement covering the Indian Army, Navy and Air Force would represent a significant step in that effort, potentially expanding the use of locally developed AI across the country’s three military services. However, the specific contract duration, scope and commercial terms could not be independently confirmed from publicly available official announcements at the time of writing.
The reported development comes as India looks to reduce its reliance on foreign technology in strategic sectors. Sarvam has already established a presence in defence-related AI through its collaboration with the Defence Services Staff College (DSSC), which introduced DRONA 2.0 in August 2026. The system combines Sarvam’s indigenous language model with other open-source models to support defence education and research, providing a documented example of how Indian-developed AI is entering military institutions.
Key Takeaways
- Reported five-year agreement: The headline describes a defence AI deal involving Sarvam and India’s Army, Navy and Air Force, although the full contract details could not be independently verified.
- Existing defence partnership: Sarvam signed an MoU with the Defence Services Staff College to develop DRONA GPT and related AI applications.
- Multimodal capabilities: DRONA 2.0 supports document analysis, image recognition, speech tools, translation and data analytics.
- Strategic objective: India is seeking to develop domestic AI capabilities for sensitive government and defence environments.
- Sovereign AI: Locally controlled models and infrastructure could help reduce dependence on foreign AI providers for certain workloads.
- Implementation remains important: Deployment, security testing, data controls and operational validation will determine how useful these systems become for military users.
What Is the Reported Sarvam Defence AI Deal?
The reported five-year arrangement would place Sarvam within a broader effort to bring AI capabilities into India’s defence ecosystem. A programme spanning the Army, Navy and Air Force could potentially support several types of work, from analysing documents and translating information to assisting with research, training and administrative processes.
However, a five-year duration alone does not reveal the agreement’s full scope. Publicly available information reviewed for this article did not establish the contract value, precise deliverables, deployment schedule or whether the arrangement covers all three services directly or through a common institutional framework.
It is also important to distinguish a technology partnership from an operational procurement contract. A memorandum of understanding may establish collaboration and development objectives without committing either party to a particular purchase volume or deployment schedule.
Until an official announcement or contract document provides further details, claims about specific operational capabilities or the number of military users should be treated cautiously.
Sarvam’s Existing Defence AI Work: DRONA 2.0
One of the clearest examples of Sarvam’s involvement in defence-related AI is DRONA 2.0, launched by the Defence Services Staff College in Wellington, Tamil Nadu, in August 2026.
According to The Times of India, the system combines Sarvam 105B with open-source language models to support education and research within the defence institution.
Its reported capabilities include:
- Document analysis and inference.
- Image recognition and generation.
- Speech-to-text and text-to-speech processing.
- Translation and transliteration.
- Research support and data analytics.
- Generating presentations and podcasts.
The system is hosted on GPU infrastructure and is tailored to the institution’s academic repository. This allows it to support military education and research using material relevant to the college’s work.
The launch was accompanied by an MoU between Sarvam and DSSC to support further development of DRONA GPT and related AI applications. This partnership is a documented milestone, although it should not automatically be interpreted as confirmation of the separate five-year, three-service agreement described in the headline.
Why India Wants Homegrown AI for Defence
Military organisations handle sensitive information, including operational documents, intelligence assessments, training material and logistics data. Sending such information to an externally controlled AI service can introduce security, governance and data-residency concerns.
Developing domestic AI capabilities gives Indian institutions more options for controlling where models run, how information is processed and who can access the underlying infrastructure.
The issue has become more prominent as AI is increasingly used for analysing large volumes of information and supporting complex decision-making. India has also been pursuing broader technological self-reliance in defence, computing infrastructure and strategic industries.
In May 2026, The Indian Express reported that the Defence Ministry was in discussions with Indian AI companies, including Sarvam and BharatGen, about integrating domestic AI models into existing defence capabilities. The report described a wider effort to reduce dependence on foreign technology in sensitive areas.
Read the Indian Express report on India’s domestic defence AI push.
Domestic development does not automatically guarantee security or superior performance. Models still require rigorous testing, secure infrastructure, access controls and human oversight, particularly when used in environments where mistakes can have serious consequences.
What Could AI Offer the Army, Navy and Air Force?
The potential applications of AI differ across military services, but several non-combat and decision-support tasks could benefit from language and multimodal models.
| Service | Potential AI applications | Key requirement |
|---|---|---|
| Indian Army | Document analysis, training support, logistics information and research | Reliable operation in restricted or disconnected environments |
| Indian Navy | Technical documentation, maintenance records, training and research | Secure access to sensitive naval information |
| Indian Air Force | Technical manuals, maintenance support, training material and information retrieval | Accurate outputs and integration with existing systems |
| Joint institutions | Translation, summarisation, knowledge management and staff research | Consistent security and governance across services |
These are potential use cases, not a confirmed list of deliverables under the reported agreement.
A language model can help users locate information, summarise long documents or translate material. More specialised systems may be needed for real-time surveillance, sensor analysis, autonomous platforms or operational control.
The distinction matters because generative AI is not a substitute for validated military systems. Any application used in a safety-critical environment would need testing against the specific requirements of that environment.
Sarvam’s Role in India’s Sovereign AI Strategy
Sarvam is an Indian AI company developing language models and supporting infrastructure for domestic applications. Its models are designed to work with Indian-language and enterprise use cases, while the company’s partnerships extend into government and strategic sectors.
In August 2026, Sarvam announced a partnership with the Centre for Development of Advanced Computing (C-DAC) to bring indigenous computing infrastructure, AI models and applications together. The stated objective is to develop an integrated domestic AI stack spanning computing hardware through to applications.
According to Sarvam’s official partnership announcement, the collaboration includes running Sarvam’s AI stack on C-DAC infrastructure and developing applications for government and strategic-sector deployments.
This is relevant to defence because model ownership is only one part of technological sovereignty. Institutions also need dependable computing resources, secure deployment environments, trained personnel and long-term technical support.
Building those capabilities domestically could give Indian organisations more control over sensitive AI workloads. Nevertheless, domestic models may still depend on imported chips or other components, so achieving full technological independence remains a broader challenge.
What the Reported Deal Could Mean for India’s AI Industry
A long-term defence engagement, if confirmed, could offer Sarvam an opportunity to demonstrate that Indian-developed models can meet the requirements of highly regulated and security-sensitive institutions.
It could also create opportunities for other domestic technology companies working on secure infrastructure, data management, specialised AI applications and system integration.
Defence environments place demands on technology that differ from ordinary consumer applications. Systems may need to work on restricted networks, operate with limited connectivity and maintain strict records of who accessed information and what actions were taken.
Successfully meeting these requirements could help domestic AI companies establish credibility in other government and enterprise markets. But the benefits would depend on the actual contract scope, deployment results and the ability to maintain reliable systems over time.
The Bigger Picture
India’s defence AI ambitions reflect a broader strategic shift: countries increasingly view AI models, computing infrastructure and data control as components of national capability. Developing local alternatives can help reduce exposure to external service restrictions and provide greater control over sensitive workloads.
Sarvam’s existing work with DSSC and its broader sovereign AI partnerships show that domestic models are beginning to enter institutional environments. The reported five-year agreement would be another potentially important development, but its exact terms need confirmation before its significance can be measured precisely.
FAQs
1. Has Sarvam signed a five-year AI deal with India’s three armed services?
The headline reports a five-year agreement covering the Army, Navy and Air Force. However, the specific contract and its terms could not be independently confirmed from publicly available official announcements reviewed for this article.
2. What is DRONA 2.0?
DRONA 2.0 is a multimodal AI system launched by the Defence Services Staff College in August 2026. It integrates Sarvam 105B with other open-source models to support defence education and research.
3. How could AI help the Indian armed forces?
Potential applications include document analysis, translation, training, logistics information, technical research and knowledge management. More sensitive operational uses would require specialised systems, extensive validation and appropriate human oversight.
4. Why does India want indigenous defence AI?
Domestic AI systems can give government and military institutions greater control over sensitive data, infrastructure and deployment decisions while reducing dependence on foreign providers for some workloads.
5. Does a defence AI partnership mean the technology is being used in combat?
No. A partnership or development agreement does not by itself establish combat deployment. The actual applications, level of operational integration and safeguards depend on the terms of the programme and subsequent implementation.
Looking Ahead
The most important next step is confirmation of the reported agreement’s scope, duration, participating institutions and deliverables. Official disclosures would help establish whether it is a development partnership, a framework for wider deployment or a contract for specific systems. Those details are necessary to assess the scale of Sarvam’s role in India’s defence AI programme
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
