/dqc/media/agency_attachments/3bO5lX4bneNNijz3HbB7.jpg)
Follow Us/dqc/media/media_files/2026/02/23/ibm-2026-02-23-12-09-21.png)
IBM, BharatGen, and NxtGen Back Swadeshi AI Stack
The Government of Andhra Pradesh has announced plans to collaborate with IBM, BharatGen and NxtGen to build a Swadeshi AI Stack. The initiative is positioned as a sovereign AI technology framework designed to enable citizen-centric AI solutions in Telugu and other regional languages.
The Swadeshi AI Stack will unify AI models, government and related datasets, a software platform and deployment tools within a single architecture. The framework is designed to operate on sovereign Cloud platforms or on-premises datacentres within Indian jurisdiction.
The stated objective is to strengthen governance and service delivery across priority sectors while building AI capabilities aligned with India’s linguistic and cultural context.
Designing a sovereign AI framework
The proposed Swadeshi AI Stack brings together four core elements:
Indigenous and domain-specific AI models
Government and institutional datasets
A unified software and deployment platform
Secure Cloud and GPU-backed infrastructure
The collaborating companies will provide technical and strategic inputs on architecture, design and operating models. The framework aims to ensure that AI workloads can be deployed with control over data, operations and governance built into the system.
The stack is intended to support compliance with localisation requirements while enabling scale across departments and sectors.
Defined roles across models and infrastructure
IBM will contribute technologies and expertise in data and AI, hybrid Cloud, governance, automation and security. The company has outlined a platform-based approach where sovereignty is embedded by design, allowing AI developers and infrastructure providers to deploy and manage workloads with operational oversight.
BharatGen will strengthen the Swadeshi AI Stack through multilingual foundation models built in India. These models provide native support for Telugu and other Indian languages and cover text, speech and vision capabilities. The foundation layer is expected to support government departments, startups, academia and industry in developing citizen-focused applications.
NxtGen will enable sovereign Cloud and GPU infrastructure optimised for AI training and inferencing. The infrastructure will be hosted in India and designed to meet data localisation, security and compliance requirements. It will support both foundational and domain-specific models.
The company has also indicated that heterogeneous GPU environments will be unified and that optimised models will be able to run efficiently on CPUs, potentially reducing the cost of delivering citizen services.
Governance vision and scale
Nara Lokesh, Minister of IT, Electronics and HRD, Government of Andhra Pradesh, said the collaboration marks a step towards building a sovereign, citizen-first AI ecosystem rooted in Indian languages and values.
Sandip Patel, MD, IBM India and South Asia, stated that indigenous models and sovereign GPU infrastructure are central to scaling government AI solutions while addressing local needs.
Prof. Ganesh Ramakrishnan, Founding Director, BharatGen Technology Foundation, said the effort will include advanced Telugu and multilingual AI models built on BharatGen’s foundational systems.
A S Rajgopal, MD and CEO, NxtGen, said sovereign AI must be supported by sovereign infrastructure and that the stack is designed to preserve data and operational sovereignty while enabling population-scale adoption.
Strategic direction
The Swadeshi AI Stack aligns with Andhra Pradesh’s broader goal of positioning itself as a leader in responsible and applied AI. The state plans to enable AI-powered hyper-personalised citizen services, foster a startup ecosystem and strengthen governance through scalable digital public infrastructure.
By integrating indigenous models with sovereign infrastructure, the Swadeshi AI Stack reflects a policy-driven approach to AI adoption that places data control, multilingual inclusion and operational resilience at its core.
Read More: