What it takes to make sovereign AI a reality for government agencies

With the geopolitical uncertainties that plague the world today, there is an increasing desire to reduce reliance on external technology providers, particularly for artificial intelligence (AI).

The shift towards technological independence and localised AI are part of what’s called AI sovereignty.

Much like national security, the ability to independently control and protect government systems and operations extends to digital systems, which is also known as exercising digital sovereignty.

Digital sovereignty is the control of digital infrastructure, technology, and data in alignment with national values, policies, and strategic objectives, as a way of safeguarding citizens’ data.

The Open Source Summit at Hyderabad, India, was a space that showed sovereign AI is no longer just an aspiration, but a possibility for governments to start building independent AI capabilities that bring more certainty to an unpredictable digital world.

To subscribe to the GovInsider bulletin, click here. 

India showing the way

The summit showcased India’s mission to advance AI capabilities in the country.

The initiative was led by the public sector with a US$1.2 billion (S$1.5 billion) fund to develop and democratise AI through open innovation.

The overall objective was to become self-reliant in AI capabilities, noted NxtGen Cloud Technologies’ Managing Director & Chief Executive Officer, Rajgopal A S.

Rajgopal noted that the current efforts served to “[provide] the infrastructure and the platform to build AI solutions for itself”.

Red Hat’s open hybrid cloud enabled a more independent approach that helped to avoid vendor lock-in and simplified migration across different infrastructures if sovereignty requirements changed.

To subscribe to the GovInsider bulletin, click here. 

Additionally, the open source model prioritises security and reliability, allowing organisations to manage their digital assets according to their specific requirements.

By providing access to sovereign infrastructure, Rajgopal hoped to see more startups and innovators testing and deploying AI solutions that served local needs.

Local talent could provide specific cultural content for AI models, for instance, by training AI models that reflect local societal norms, values, and language.

Rajgopal shared that to realise this vision of self-reliance, SpeedCloud relied on Red Hat’s OpenShift as the unifying platform to “allow the customer to focus on his own AI use case”.

This unification of different software enables SpeedCloud to create the foundation for a platform-agnostic cloud.

This flexibility was an essential aspect of digital sovereignty as it eliminated potential vendor lock-in and maintained a seamless experience across environments.

Red Hat’s Chief Technology Officer for APAC, Vincent Caldeira, noted that India’s open-source-first approach “offered a new paradigm for AI sovereignty that balances national control with global collaboration, serving as a powerful model for other nations”.

Securing local talent through sovereign AI

AI sovereignty not only serves to enhance protection for data privacy but could also boost economic competitiveness.

By fostering an AI innovation ecosystem that encourages AI experimentation and development, local talents and enterprises have more opportunities to deliver value beyond proof of concepts.

In Canada, for example, the Sovereign AI Compute Strategy launched in 2024 aimed to guide investment to support the local AI ecosystem.

The government is expected to invest up to CD$700 million (S$650 million) to support Canada’s AI compute capacity, and up to CD$1 billion (S$929 million) to build sovereign public AI infrastructure.

Canada also recently launched a public consultation to understand the focused areas for the government to accelerate safe adoption of AI, scale local talent and strengthen sovereign infrastructure.

Other countries, like Australia, are starting to look at the risks and opportunities that come with implementing AI in government.

These considerations are opening the door to explore more intentional pathways toward achieving true AI sovereignty.

Open source AI models, for instance, stand as a key enabler for transparency and control by providing visibility of the code and the logic that drives AI outputs, minimising risks of data exposure that come with proprietary LLMs.

This technology sovereignty also involves new opportunities for talent development to ensure that AI systems are handled by local talent with the necessary skills.

What it takes to succeed

Incorporating open source can help make sovereign AI a reality for government agencies.

India’s case showcased the potential of open source in making sovereign AI a reality. Malaysia’s AI ecosystem also sees open source as a critical leveller by lowering costs, enabling autonomy, and offering flexibility to local needs.

Governments can tap on these capabilities to achieve self-reliance and deploy AI on their own terms.

To subscribe to the GovInsider bulletin, click here. 

Canada’s efforts showed that to support a successful move toward sovereign AI and reap its benefits, a committed investment toward strengthening a digital infrastructure was critical.

Beyond the resources, clear governance was also key to ensure that the innovations benefit the public while adhering to both national and international standards.

Last August, speakers at the ASEAN AI Summit in Kuala Lumpur, Malaysia, noted that to advance AI in the region, local efforts to improve domestic AI governance were a critical first step.

A white paper presented at the Summit, on Considerations regarding Sovereign AI and National AI Policy, pointed out that “sovereign AI is not a binary goal; it exists on a spectrum of strategic choices, where nations must balance self-reliance with global AI interoperability”.    

Open source for a sovereign future    

The impact of effective AI deployment was also projected to generate 50 per cent of new economic value from digital businesses in Asia Pacific and Japan by 2030, according to IDC.

With over 460 million digital users in the Southeast Asia, region alone, countries are now balancing the vision of pursuing domestic AI advancement individually with cooperation efforts to achieve harmonised standards for transparency.

The time to move toward sovereign AI may be more fitting now with so much potential at hand for governments to scale their AI capabilities.

Beyond clear standards for governance and a stable stream of resources, the adoption of secure, flexible open source platforms like Red Hat OpenShift Platform Plus can help to embed digital sovereignty and autonomy for organisations that are starting to make this move.

An open source-first approach can thus serve as the foundation for building a resilient and responsible AI ecosystem for government agencies to accelerate AI adoption and enable local talent growth.