The public sector must act as anchor tenant, SMEs must be given support to adopt AI and all workers must take to it.

By Selena Ling, Chief Economist and Head of OCBC Group Research

When Singapore talks about becoming an artificial intelligence (AI) hub, it is tempting to frame the ambition as a familiar story. We have done this before, after all.

In the 1990s, we built wafer fabrication plants and embedded ourselves into global semiconductor supply chains. In the 2000s, we wired the island with broadband and rolled out e-government. Each time, the state moved early, coordinated tightly, and executed efficiently.

But AI is not another wafer fab. Nor is it just digitalisation 2.0. It is something more systemic, more diffuse and more destabilising.

The question is not whether Singapore can launch an AI strategy. It already has. The harder question is whether it can build an AI ecosystem deep enough, broad enough and trusted enough to generate sustained economic transformation – and whether a top-down National AI Council will be enough to make it happen.

To answer that, it helps to look at what has worked elsewhere and what has not.

What an AI hub actually means

Being an AI hub is not about having a handful of tech firms or flashy announcements about compute clusters. The question is how Singapore can move from early leader to entrenched, trusted anchor in a fragmented global market. It requires five reinforcing elements:

First, frontier research capability. Second, a dense and renewable talent pipeline. Third, deep pools of risk capital. Fourth, widespread industry adoption beyond tech firms. Fifth, public trust in how AI is deployed. Miss one, and the flywheel slows.

Singapore’s strength historically lies in coordination and infrastructure. But AI hubs are not built on infrastructure alone. They are built on ecosystems. Consider the example of Silicon Valley. Its advantage was not government coordination. It was network density – venture capitalists, university labs, serial entrepreneurs and corporate giants reinforcing one another. Talent flows easily, and failure is tolerated. As risk capital is abundant, ideas can flourish and compound.

Now contrast that with Beijing, which became a major AI centre through state alignment with industrial policy. National AI plans channelled funding into strategic sectors, while domestic tech giants provided scale and data. Government direction mattered. But so did a vast domestic market and aggressive capital deployment.

Singapore resembles neither fully. It does not have Silicon Valley’s risk culture nor Beijing’s scale. Its model must therefore be hybrid – coordinated but porous, strategic yet open.

Uncharted territory

When Singapore pursued semiconductors, the value proposition was clear: provide land, utilities, skilled engineers and political stability. Multinationals would anchor production here. Capital intensity was high, but once established, operations were sticky.

AI does not work that way. Its most valuable inputs are mobile: data, algorithms and people. Talent can relocate quickly. Code can be deployed globally overnight. Competitive advantage potentially erodes faster than you can blink an eye.

Moreover, AI’s impact is cross-sectoral. It will shape finance, logistics, healthcare, manufacturing, education and even public administration. That makes diffusion – not just concentration – the real test of success.

If only elite tech firms benefit, productivity gains will remain narrow. But if small and medium-sized enterprises (SMEs) integrate AI into pricing, inventory, customer analytics and operations, the effect multiplies. This is where past digitalisation efforts offer both lessons and warnings. Programmes under Singapore’s Smart Nation push improved government services significantly, but SME adoption has often lagged beyond basic digitisation. AI adoption will be harder still – it requires structured data, cultural change and process redesign.

The case for and limits of a National AI Council

Singapore’s instinct to coordinate at the centre is understandable. AI policy cuts across trade, education, manpower, finance and national security. A high-level council can align incentives and prevent fragmentation. But councils do not innovate, people do. The risk of a purely top-down model is bureaucratic over-optimisation – safe bets, incremental grants, excessive reporting requirements. Innovation ecosystems require tolerance for failure and uneven outcomes.

A successful AI council must therefore do three things: set clear strategic niches rather than chase everything; ensure regulatory clarity without over-prescription; empower bottom-up experimentation, including from start-ups.

If it becomes merely another inter-agency committee, momentum will stall. At the end of the day, will chief executive officers say “Singapore = AI” like they say “Changi = airport”?

Spreading benefits across the economy

AI’s promise lies in productivity. For diffusion to occur, three conditions are necessary.

First, talent must extend beyond data scientists. Lawyers, accountants, nurses, port operators and teachers need AI literacy. Without broad capability, adoption remains shallow. This may also require an overhaul of education end-to-end, from primary school AI literacy to SkillsFuture pathways mapped by occupation which may currently be too fragmented.

Second, SMEs need structured support. Templates, shared models, subsidised experimentation and advisory networks must reduce the cost of first adoption. AI cannot remain confined to large financial institutions or global tech players. Without diffusion, a two-tier economy potentially emerges: multinational corporations thrive, while smaller players stall.

Third, the public sector must act as anchor client. Government procurement of local AI solutions can provide early revenue and credibility, much as defence and aerospace contracts did in other economies.

If AI becomes visible in healthcare triage systems, transport optimisation and housing estate management, trust and familiarity will grow.

What would success or failure look like?

In the first three years, the visible signs will be investment announcements and talent inflows. That is the easy part.

Between years three and seven, the harder metrics should emerge. These would include rising AI start-up formation and venture funding depth; measurable productivity gains in AI-intensive sectors; export revenue from AI-enabled services and international research collaborations and citations.

Beyond seven years, structural effects should appear: stronger total factor productivity growth and higher-value job creation.

Failure, by contrast, would be quieter: Talent attrition, under-utilisation of compute infrastructure, unconvinced SMEs, and some erosion of public trust. The greatest risk is not that Singapore does nothing. It is that it does something visible, but that is insufficiently embedded.

Managing inequality and trust

AI will disproportionately reward high-skill workers at first. If not managed carefully, income dispersion may widen. Reskilling frameworks, wage support mechanisms and transparent communication are therefore not social add-ons, but economic prerequisites. Public backlash can derail technological progress, as seen in other jurisdictions where algorithmic bias or data breaches triggered distrust.

Trust therefore is not a soft variable, but a competitive advantage. Singapore may not match Silicon Valley in frontier breakthroughs. Nor can it rival Beijing in scale. But it can differentiate itself in trusted AI governance – systems that are auditable, secure and exportable. That niche would align naturally with Singapore’s reputation as a financial and regulatory centre.

How long will Singapore’s AI hub ambitions take to materialise? Realistically, up to a decade? Industrial policy cycles are long and ecosystems tend to mature slowly. The semiconductor push took decades to embed deeply. Digital transformation spanned a generation. AI will move faster technologically, but institutional transformation remains gradual. The first visible wins may arrive within five years. The true test will be whether AI measurably lifts productivity growth in the 2030s.

AI as a strategic choice

Ultimately, the ambition to become an AI hub is less about prestige than about economic survival. Small, open economies cannot compete on scale. They compete on adaptability. AI is not merely a sector to develop. It is a general-purpose technology that will redefine sectors. The question is whether Singapore shapes that shift or reacts to it.

A National AI Council can set direction to enable regulation and catalyse funding, but the decisive variable will be whether AI becomes embedded – not just in research labs or large banks – but across the everyday operating systems of the economy. If it does, Singapore will not simply be an AI hub. It will be an AI-enabled nation. If it does not, ambition will outpace impact. And in the AI era, the distance between the two can widen quickly.

The 2026 Council is a defensive-offensive move to lock in gains before geopolitics or talent caps stall momentum. Success indicators will be less about gross domestic product and more about whether a local SME can productise an AI agent and whether a 70-year-old uses GovGPT without knowing it.

This article was first published in The Straits Times on 14 February 2026.