Neuralix provides a defence-focused foundational model and a sovereign GenAI platform operating inside the Indian Army’s isolated, air-gapped network.
“India’s next generation of military AI must be sovereign by design. Neuralix has demonstrated that this capability can be built from the ground up, securely and at mission speed.”— Maj Gen Ravi Murugan, PVSM, AVSM, Senior Adviser, Neuralix AINEW DELHI, INDIA, September 2, 2026 /EINPresswire.com/ — Neuralix AI Private Limited has created a sovereign AI foundational model that was built entirely from scratch specifically for military use, along with a unified Generative AI platform that functions completely inside the Indian Army’s secure digital environment. This accomplishment marks a significant step forward in sovereign defence artificial intelligence. Instead of modifying a general-purpose or open-source model for military tasks, Neuralix built a foundational model centred on defence needs, featuring its own architecture and model weights, as well as a training and development pipeline tailored to the military domain.
The model was built to grasp defence terminology, military concepts, doctrine, strategy, complex Army documents, and domain-specific knowledge, allowing it to produce structured insights for authorised military personnel. Its knowledge and retrieval framework was likewise designed around military information needs rather than simply applying conventional retrieval-augmented generation to a generic language model. The foundational model employs a Mixture-of-Experts architecture and was developed inside the Indian Army’s environment, ensuring that sensitive data stays under Army control throughout both development and operation. Neuralix’s efforts are being conducted under the Indian Army’s AI-as-a-Service with infrastructure setup challenge as part of the iDEX ADITI 2.0 programme. Neuralix was chosen as the technology developer following a competitive process involving several candidates. The Directorate General of Information Systems (DGIS) acts as the Indian Army’s steering and user agency, Neuralix as the technology developer, and the Defence Innovation Organisation as the contracting agency under iDEX ADITI 2.0.
In addition to the foundational model, Neuralix has developed EKAM, a unified sovereign military Generative AI platform meant to operate entirely within the Indian Army’s digital network and compute environment. EKAM was introduced by the Hon’ble Raksha Mantri during the Chanakya Defence Dialogue and offers a secure platform capable of hosting multiple AI models, specialised services, and military use cases. EKAM functions within an air-gapped environment, cut off from the public internet and external cloud services. It can support the military foundational model as well as specialised Small Language Models created for specific Army needs and other AI capabilities that may be added in the future. Together, the platform and foundational model establish sovereignty across three critical layers: the model, the data, and the infrastructure. Model sovereignty means that the AI system’s architecture, weights, and evolution can be kept in controlled custody rather than depending on an external commercial AI provider. Data sovereignty guarantees that confidential Army information remains within the Army environment. Infrastructure sovereignty allows the entire system to operate on Army-controlled compute and networks without reliance on public cloud infrastructure or external APIs.
Building modern AI inside such an environment posed a significant engineering challenge. Most contemporary AI development assumes constant access to software repositories, model hubs, external libraries, APIs, and cloud services. Within an air-gapped defence network, these dependencies cannot simply be accessed on demand. Neuralix engineers had to securely bring required dependencies into the environment, modify components where necessary, and in some cases develop or adapt software specifically so that the complete AI stack could function efficiently without external connectivity. Neuralix employed a forward-deployed engineering approach throughout the programme, with its technical team working inside the environment provided by DGIS and building the technology closely around Army requirements.
This approach allowed the technology to be built around the user while confidential Army data remained with the user. Sensitive information did not need to be transferred to an external AI environment for Neuralix to develop, test, and refine the system. In just over a year, Neuralix has developed both a unified sovereign military GenAI platform and a ground-up military foundational model within a highly secure and restrictive defence environment. The two capabilities represent distinct but complementary achievements. EKAM provides the sovereign platform and infrastructure layer, while the military foundational model offers an intelligence layer designed specifically around defence knowledge and military requirements.
Over time, this architecture could serve as the foundation for a broader military AI ecosystem, potentially supporting specialised military agents, decision-support systems, operational software, edge-deployed AI capabilities, and other secure applications developed around future Army requirements. Any expansion into additional operational or military systems would be subject to the appropriate development, evaluation, and deployment processes. For Neuralix, the programme demonstrates that an Indian technology company can build a military-specific foundational model from the ground up, create a complete sovereign GenAI platform around it, and make the full technology stack operate within an air-gapped defence environment while keeping sensitive military data under Army control.
As governments worldwide evaluate how artificial intelligence should be deployed in defence and other critical national infrastructure, control over models, data, and infrastructure is becoming increasingly important. Neuralix’s work with the Indian Army illustrates one approach to bringing these requirements together: building the model for the mission, building the technology around the user, and keeping sovereignty at the foundation of the entire AI stack.
Tarang Singh
Neuralix AI Pvt. Ltd.
info@neuralix.ai
