Small states, big choices: Singapore’s approach to sovereignty in the age of AI

AT THE AI Impact Summit from Feb 16 to 20 in New Delhi, a central question emerged for governments: How can states harness artificial intelligence (AI) at scale while remaining open, competitive, and sovereign in a deeply interconnected world?

Too often, however, AI sovereignty is framed as a race for self-reliance – who owns the most data, builds the largest models, or controls the most compute. For small, open economies like Singapore, this framing isn’t just unrealistic, it risks obscuring what sovereignty actually requires.

AI sovereignty is best understood as a set of strategic postures, not a quest for technological self-sufficiency.

Different countries will approach it from different starting points and priorities. Large economies may seek greater control over foundational models and compute. Resource-rich states may prioritise domestic infrastructure. Smaller or more open economies face a different strategic reality.

For these states, AI sovereignty has never meant owning every layer of the system. It has meant retaining the ability to choose, adapt, and act in a rapidly-changing geopolitical environment.

Tony Blair Institute’s Sovereignty in the Age of AI framework reframes sovereignty not as a binary choice between control and dependence, but as a spectrum of strategic postures across different layers of the AI stack.

This lens helps explain why countries may exercise direct control in some areas, steer through standards and partnerships in others, and manage dependence where scale advantages lie elsewhere.

Agency over ownership

In the AI era, agency matters more than ownership. The ability to make deliberate, future-oriented decisions about how AI is integrated into the economy and governed across society will shape national outcomes far more than whether a country builds its own frontier models.

The strategic question is not who owns the technology, but who decides how it is deployed, under what conditions, and to whose benefit. Sovereignty, in this sense, is exercised through deliberate choice, not isolation.

Singapore’s approach reflects this distinction clearly. Instead of pursuing full self-reliance in AI infrastructure, the country partners with global hyperscalers for compute while maintaining domestic high-performance computing capabilities for research and experimentation. This allows Singapore to preserve agency where it matters, without incurring the costs and constraints of trying to replicate global scale.

This logic is not new to Singapore. Its approach to AI reflects a long-standing way of treating sovereignty: practical, selective and forward-looking – the same logic that shaped its trade, energy and digital strategies.

Rather than attempting to out-research the United States or out-deploy China, Singapore focuses on building the judgment capability to decide when and how AI should be deployed. Sovereignty is grounded in human decision-making and institutional capacity, not technological nationalism.

Investing in people

Talent sits at the heart of Singapore’s approach. People – not compute or models – are the hardest capability to build quickly, yet the easiest to lose if neglected. The city-state identified talent early as a sovereignty lever to meaningfully build and strengthen.

Through programmes such as AI Singapore’s AI Apprenticeship Programme (AIAP), over 500 AI engineers have been trained domestically, with around 95 per cent placed into industry – embedding AI capability directly into the economy. This investment is treated as a form of state capacity, not just manpower policy.

Flagship platforms such as LearnAI and AIAP were designed as sustained national pipelines – from students to mid-career professionals – aligned to evolving needs. Over time, this investment has yielded a pool of practitioners capable of deploying AI responsibly across wide-ranging domains, from healthcare and finance to logistics and urban systems. For Singapore, talent development has become a way to anchor sovereignty in people rather than infrastructure alone.

Driving real-world adoption

Singapore’s emphasis on applications over models further illustrates this point. Real value from AI is realised through use, not architecture. Rather than competing in foundational model development, Singapore deliberately focuses on solving real problems in the public and private sectors.

Programmes like the 100 Experiments programme, Short Industry Projects, and LLM Applications Developer Programme, have supported more than 300 real-world projects, with over half progressing to deployment – showing that adoption is the key to driving scalable and lasting impact.

Moreover, an application-first strategy allows smaller states to punch above their weight. By prioritising deployment and institutional adoption, Singapore has extracted outsized value from global AI innovation without needing to dominate upstream research. Here, openness becomes an advantage rather than a vulnerability.

Sovereignty is also strengthened through selective control and managed inter-dependence. Singapore exercises strongest control where leverage is highest – talent, applications and governance – while partnering globally in hardware-intensive layers where scale advantages lie elsewhere. Regional models such as Sea-Lion – a multimodal AI model built to understand South-east Asia’s languages and contexts – illustrate a strategy of fine-tuning and adaptation rather than building from scratch.

Singapore has extended its influence through governance leadership – developing tools such as AI Verify, participating actively in ISO/IEC standards, and co-leading global conversations through global forums like Global Partnership on AI. These efforts allow a small state to shape norms beyond its borders even where it does not dominate the underlying technologies.

Sovereignty as a strategic design choice

Crucially, the lesson here is not for other countries to replicate Singapore’s approach wholesale, but to treat sovereignty as a strategic design choice. By framing sovereignty as a set of strategic choices across different layers of the AI stack – rather than a single ambition for self-sufficiency – it offers governments a way to think clearly about trade-offs, constraints and priorities. Each country can chart its own path, informed by its economic structure, institutional strengths and geopolitical context.

Singapore’s experience illustrates how such choices can be made deliberately and consistently in practice. As AI continues to reshape the world economy, governments that approach sovereignty strategically – rather than rhetorically – will be best positioned to build long-term resilience, sustain economic competitiveness, and preserve agency in a deeply interconnected world.

Tay PeiChin is senior policy advisor (government innovation) at the Tony Blair Institute for Global Change and Laurence Liew is director of AI Innovation at AI Singapore