The Indian Navy’s Weapons Electronics Systems Engineering Establishment originally laid the groundwork for this evolution through the launch of the initial DRONA gateway in 2025. This technological foundation served as a vital stepping stone toward the more robust, multifaceted intelligence system that has now arrived within the halls of the Defence Services Staff College in Wellington. On August 18, 2026, the military officially introduced DRONA 2.0, also known as the DSSC Resource Optimising Neural AI, marking a historic pivot toward a multimodal generative future. This platform is not merely an upgrade but a complete reimagining of how high-level officers interact with data, moving beyond simple keyword searches to complex, context-aware analysis. By integrating the indigenous Sarvam 105B model, the Indian Armed Forces established a precedent for technological sovereignty that ensures operational secrets remain protected within a framework designed specifically for the unique rigors of national defense.
Technological Infrastructure and Security Protocols
The Architectural Foundation of the Sarvam 105B Model
At the heart of this advancement lies the Sarvam 105B model, a specialized indigenous Mixture-of-Experts (MoE) architecture that was born out of the national IndiaAI Mission. The MoE structure is particularly advantageous for military applications because it does not activate the entire neural network for every single query. Instead, the system identifies the specific sub-networks most qualified to handle a particular task, whether it involves linguistic translation or heavy statistical computation. This selective activation ensures that the system remains incredibly fast and energy-efficient, even when processing massive datasets. For an officer working under time constraints to draft a strategic briefing, the speed and accuracy provided by this architecture are indispensable. The model is also uniquely tuned to the linguistic subtleties of the Indian context, allowing it to interpret localized terminology and cultural contexts that foreign-built general-purpose models would likely overlook.
This technical foundation provides the Indian military with a level of transparency and auditability that is impossible to achieve with proprietary foreign models. Because the Sarvam 105B was engineered domestically, national authorities have full access to its underlying weights, biases, and training methodologies. This transparency is critical for maintaining the ethical and legal standards required in military decision-making. It allows for rigorous testing against adversarial attacks and ensures that the AI’s reasoning can be deconstructed and verified by human experts. The Mixture-of-Experts approach also allows for modular updates; as new military doctrines are developed or new threats emerge, specific experts within the model can be retrained or expanded without needing to overhaul the entire system. This flexibility ensures that the platform remains cutting-edge and highly adaptable to the shifting landscape of global security, providing a robust and scalable solution for national personnel.
Sovereign Security and Operational Applications
Recognizing that military data is a prime target for foreign espionage, the architects of DRONA 2.0 implemented a stringent on-premise hosting strategy. The system is housed on high-performance GPU servers located physically within the secure, internal network of the Defence Services Staff College. This air-gapped approach ensures that no sensitive data ever leaves the premises or traverses the public internet, effectively neutralizing the risk of cloud-based surveillance or remote data breaches. For mid-level officers conducting sensitive research into regional security or tactical vulnerabilities, this localized deployment provides an essential layer of psychological and operational security. They can engage with the AI’s most advanced features without fearing that their prompts or the system’s generated responses will be intercepted by unauthorized parties. This focus on physical and digital containment reflects a mature understanding of the necessity of preserving data sovereignty.
To ensure this technological advantage remained sustainable, the defense establishment focused on creating a long-term roadmap for continuous improvement and integration. The successful deployment at Wellington served as a catalyst for expanding the platform’s reach to other specialized training centers and operational commands across the country. Future efforts were directed toward refining the model’s reasoning capabilities through the incorporation of real-time battlefield data and expanding its linguistic database to include more obscure regional dialects. Decision-makers recognized that the value of such a system grew in proportion to the quality of the data it processed, leading to a renewed emphasis on secure data collection and categorization. By maintaining a close partnership with indigenous tech firms, the military ensured that the platform remained at the cutting edge of global innovation. This proactive approach guaranteed that India’s leaders remained prepared for future shifts in modern warfare.
