Sovereign by design - Building India’s independent, inclusive and future-ready AI

Sovereign AI is the ability to develop, deploy and govern artificial intelligence on India’s own terms. It is about control, continuity, public value and national priorities. It does not mean building every technology independently. Instead, India needs enough control to protect essential systems while still benefiting from global innovation.

That control covers critical data, models, compute and rules. It also includes infrastructure, procurement, governance and talent. A sovereign system should be designed for data residency, portability and multi-model flexibility. The government must define sovereignty by layer and track progress with measurable indicators.

India has already built public AI capacity at speed. The next challenge is making it trusted, scalable and durable. The report proposes governed data exchanges, procurement reform and funding for ongoing operations. By 2030, it envisages population-scale AI services, sovereign-controlled models for priority services and independent assurance.

The six foundational layers are compute; data; models; platforms and digital public infrastructure; talent and intellectual property; and governance and trust. Together, they describe the capabilities India needs to build, operate and oversee AI systems. The framework avoids treating sovereignty as a single technology problem.

Compute provides the infrastructure needed to run AI. Data supports development and use. Models perform AI tasks. Platforms and DPI help deliver services at scale. Talent and IP support local capability and ownership. Governance and trust provide evaluation, transparency, accountability, human oversight and responsible use.

The report says each layer requires a distinct approach to capability development and control. This matters because weaknesses in one layer can limit the value of the others. India’s next step is to define sovereignty across these layers and measure progress through shared national milestones toward 2030.

India’s sovereign AI plan builds on its experience creating public digital infrastructure. Aadhaar, UPI, DigiLocker and Bhashini are named as examples of systems that helped establish shared foundations for digital services. The report applies that model to AI, combining public infrastructure with open innovation.

The key mechanism is a common foundation that many users and builders can access. For AI, this means developing public capabilities across compute, data, models, platforms and DPI, talent and IP, and governance and trust. It also means creating governed data exchanges and encouraging partnerships for scale.

This approach can move AI beyond isolated demonstrations. The report calls for AI to become embedded in budgets, processes, people and policy. It envisages population-scale public services across all scheduled languages, supported by sovereign-controlled models for priority services and independent assurance.

The report frames control as essential because AI may become part of public services, budgets, processes and policy. If India cannot control critical data, models, compute or rules, it may not be able to ensure continuity or protect public value. Its ability to pursue national priorities could also become constrained.

The practical risk is dependence without flexibility. Systems may be difficult to move, evaluate or govern if infrastructure, models or data arrangements are not designed for sovereignty. The report therefore highlights data residency, portability, multi-model flexibility, independent assurance and human oversight as important design choices.

India has built public AI capacity quickly, but the next four years will determine whether it can control, scale and trust that capacity. The report calls for sovereignty to be defined by layer, measured through indicators and supported by procurement reform and ongoing operational funding.

The report argues that AI growth must be inclusive by design. India’s population spans many languages, geographies and socio-economic segments. Multilingual, voice-first and last-mile systems can help public services reach people who may not be well served by narrow, text-heavy or centrally designed tools.

The mechanism is design for actual use across communities. AI services should operate in all scheduled languages and use voice where that is more accessible. Last-mile delivery focuses attention on citizens beyond the places and groups most easily reached by digital systems. Bhashini is part of the public infrastructure experience informing this approach.

The report’s 2030 vision is AI-enabled public services operating at population scale across all scheduled languages. Reaching it requires trusted governance, suitable infrastructure and inclusive implementation. Industry, startups, academia and government each have roles in building and scaling these systems.

The report makes a clear distinction between sovereignty and complete self-sufficiency. India does not need to build every AI component independently. Sovereign AI means having enough control over critical parts of the system to develop, deploy and govern AI on India’s own terms.

That approach leaves room for global companies and innovations. The important safeguards are control over critical data, models, compute and rules, along with data residency, portability and multi-model flexibility. These choices help India benefit from outside technology without becoming unable to switch or protect public value.

The report presents sovereignty as deliberate, measurable choices about who controls the data, models, infrastructure and ability to switch. Government must define sovereignty by layer. Industry must architect for flexibility, while startups and academia contribute research, benchmarks and partnerships that strengthen India’s broader ecosystem.

Data is the information AI uses, while models are the systems that perform AI tasks. Compute is the processing infrastructure that runs them. Platforms and digital public infrastructure help deliver AI services at scale. These layers determine whether an AI system can work reliably and reach users.

Talent provides the people who build, operate and improve the system. Intellectual property helps protect and grow innovation. Governance and trust cover evaluation, transparency, accountability, human oversight and responsible AI. Together, these mechanisms make systems more dependable and more acceptable for public use.

The report groups the layers into a sovereign AI stack: compute; data; models; platforms and DPI; talent and IP; and governance and trust. Each requires a distinct approach to capability and control. India’s future plans therefore need shared milestones, governed data exchanges and institutions that can sustain AI beyond pilots.