Neuralix builds sovereign military AI model for Indian Army

Neuralix AI Private Limited has developed a ground-up military foundation model and sovereign GenAI platform inside the Indian Army’s secure, air-gapped environment. The project, completed under iDEX ADITI 2.0, is designed to keep model, data and infrastructure control inside Army networks as defence organizations push for more secure AI.

Why it matters: - Neuralix’s work shows how defense AI can be built without depending on external cloud services, public internet access or commercial model providers. - The project keeps sensitive Army data, model weights and infrastructure inside the Indian Army environment. - The approach could become a template for future military AI systems, including decision-support tools, specialized agents and edge deployments.

What happened: - Neuralix AI Private Limited developed a sovereign AI foundation model built ground-up for military use. - The company also built EKAM, a unified sovereign Generative AI platform for the Indian Army. - Both systems operate inside the Indian Army’s secure, air-gapped digital environment. - The work was carried out under the Indian Army’s AI-as-a-Service with infrastructure setup challenge under the iDEX ADITI 2.0 programme. - Neuralix was selected through a competitive process with multiple contenders. - The Directorate General of Information Systems, or DGIS, is the Indian Army’s steering and user agency. - The Defence Innovation Organisation is the contracting agency under iDEX ADITI 2.0.

The details: - The foundation model was designed around defence requirements instead of adapting a general-purpose or open model. - Neuralix built the model with its own architecture and model weights. - The model is designed to understand defence terminology, military concepts, doctrine, strategy, complex Army documents and domain-specific knowledge. - The system is designed to generate structured insights for authorized military users. - The knowledge and retrieval architecture was built for military information requirements rather than standard retrieval-augmented generation on a generic language model. - The foundation model uses a Mixture-of-Experts architecture. - Neuralix developed the model within the Indian Army environment, keeping sensitive data under Army control during development and operation. - EKAM was launched by the Hon’ble Raksha Mantri during the Chanakya Defence Dialogue. - EKAM can host multiple AI models, specialized services and military use cases. - EKAM can support the military foundation model and specialized Small Language Models built for specific Army needs. - EKAM operates entirely within the Army’s digital network and compute environment. - EKAM is isolated from the public internet and external cloud services. - The platform and model together establish sovereignty across three layers: model, data and infrastructure. - Model sovereignty means the architecture, weights and evolution of the AI system remain under controlled custody. - Data sovereignty means confidential Army information stays inside the Army environment. - Infrastructure sovereignty means the full system runs on Army-controlled compute and networks without public cloud dependence. - Most modern AI development assumes access to software repositories, model hubs, external libraries, APIs and cloud services. - Neuralix had to securely bring dependencies into the air-gapped environment, modify components and adapt software so the full AI stack could run without external connectivity. - Neuralix used a forward-deployed engineering approach, with its technical team working within the environment provided by DGIS. - The company built and refined the system without moving sensitive Army data to an external AI environment. - Neuralix says it completed both the sovereign GenAI platform and the military foundation model in just over a year.

Between the lines: - The project reflects a broader shift in defense AI toward control over data, models and infrastructure rather than speed alone. - Building inside an air-gapped military network is technically harder than building on open cloud infrastructure, but it reduces exposure of sensitive information. - Neuralix’s result suggests Indian companies can develop defense-grade AI systems end to end without relying on foreign or external AI stacks. - The architecture separates the platform layer from the intelligence layer, which could make future upgrades or new military applications easier to add.

What's next: - Neuralix says the architecture could support a wider military AI ecosystem over time. - Potential future uses include specialized military agents, decision-support systems, operational software and edge-deployed AI tools. - Any extension into additional operational or military systems would go through the appropriate development, evaluation and deployment processes. - The company says the programme demonstrates how sovereign AI can be built for defense and other critical national infrastructure needs.

The bottom line: - Neuralix has delivered a military-specific AI stack for the Indian Army that keeps the model, data and infrastructure inside a secure defense environment.