▶️ LISTEN TO THIS ARTICLE

In December 2024, Singapore scored 84.25 on Oxford Insights' Government AI Readiness Index, second only to the United States at 87.03. But dig into the sub-scores and the picture flips: Singapore ranked first globally in both the Government pillar (90.96 vs. the US's 89.26) and the Data and Infrastructure pillar (93.14 vs. 90.90). A nation of 5.9 million people is outscoring every major power on government AI implementation. That gap tells a story about what happens when a small country decides AI governance is a competitive advantage, not a compliance burden.

The S$1 Billion Bet

Singapore's National AI Strategy 2.0 (NAIS 2.0), announced in late 2023 by then-Deputy Prime Minister Lawrence Wong, set targets that would be ambitious for countries ten times its size: triple the AI practitioner workforce to 15,000, establish the city-state as a global hub for AI creators, and execute 15 courses of action over three to five years.

The money followed. In 2025, the government committed over S$1 billion across five years under the National AI Research and Development (NAIRD) Plan, drawn from the National Research Foundation's S$37 billion research budget unveiled in December 2025. NAIRD focuses on three areas: fundamental AI research (including AI safety), industry partnerships for real-world deployment, and talent pipelines through programs like the AI Singapore PhD Fellowship and the AI Accelerated Masters Program.

That talent pipeline matters because Singapore can't compete on headcount. Instead, the NAIRD plan aims to nurture what it calls "bilingual research talents" with deep AI expertise and equally deep domain knowledge in fields like healthcare, finance, and logistics. At the pre-university level, the National Olympiad in AI prepares students for international competition. At the graduate level, the AI Singapore PhD Fellowship Program and the AI Accelerated Masters Program are being scaled up to feed the workforce target. The February 2026 national budget doubled down with AI-centered tax breaks and support measures for companies adopting AI tools, signaling that the government sees AI fluency as a workforce-wide priority, not just a specialist skill.

AI Verify: Governance You Can Actually Test

Most countries write AI principles documents. Singapore built software.

Most countries write AI principles documents. Singapore built software. AI Verify, launched as a minimum viable product in May 2022 by the Infocomm Media Development Authority (IMDA), is an open-source testing framework that lets organizations validate their AI systems against 11 internationally recognized governance principles, covering transparency, explainability, fairness, robustness, and data governance.

The toolkit was open-sourced on GitHub in June 2023, and the AI Verify Foundation now has over 90 member organizations. In February 2025, the Foundation and IMDA launched the Global AI Assurance Pilot at the Global AI Summit in France, pairing 16 AI testing firms with 17 companies across 10 sectors including finance, healthcare, and public services. Between March and May 2025, a coalition of over 100 participants from 30+ organizations conducted the world's first systematic technical testing of real-world generative AI applications.

The pilot's key finding was practical, not theoretical: GenAI risks are highly context-dependent. The same model behaves differently across use cases, industries, cultures, and languages. That insight matters because it undermines the idea that any single regulatory framework can cover all AI risk, a finding that adds nuance to the global AI safety debate. Singapore's response has been to align AI Verify with the OECD AI Principles, the GPAI Code of Practice, and EU/UK/US assurance models, creating interoperability rather than competing standards.

The Compute Infrastructure Race

Governance frameworks don't run on good intentions. They need hardware. Singapore has been building AI compute capacity at a pace that caught even the chip industry off guard. In fiscal Q3 2024, Singapore accounted for roughly 15% of Nvidia's global revenue, approximately $2.7 billion, making it Nvidia's fourth-largest market worldwide.

Singtel partnered with Nvidia in early 2024 to launch GPU-cloud services for Southeast Asia, with an eight-story, 58MW data center in Singapore scheduled to go online in 2025 offering Nvidia Hopper architecture GPUs. Sustainable Metal Cloud operates H100 clusters with up to 2,048 GPUs per cluster across two Singapore availability zones, with H200 deployments planned for late 2025. The government separately committed S$270 million for a next-generation supercomputer integrating classical and quantum computing capabilities.

This infrastructure build serves a dual purpose. It makes Singapore a credible AI research hub, and it positions the city-state as the compute gateway for a region where AI could add $1 trillion to GDP by 2030. According to the 2024 Google, Temasek, and Bain & Company e-Conomy SEA report, Southeast Asia attracted over $30 billion in AI infrastructure investment in the first half of 2024 alone, with tech giants including Microsoft, Google, and Amazon committing $50 billion to the region's AI sector since early 2023. Singapore sits at the center of that capital flow, offering what companies need: reliable power, English-speaking workforce, data protection laws, and proximity to the 680 million consumers across ASEAN. The UAE plays a similar gateway role for the Middle East, though with a very different governance philosophy.

Pick a layer of the AI stack where size doesn't matter, build practical tools, and open-source everything so adoption scales beyond your borders.

The Risks in the Model

Singapore's approach isn't without vulnerabilities. The US investigation into alleged GPU transshipments to China through Singapore highlights the geopolitical tightrope the city-state walks. Being a chip trade hub means being subject to great-power scrutiny, and any perception that Singapore is a backdoor for restricted technology could threaten its access to the very hardware its AI strategy depends on.

There's also the talent constraint. Tripling the AI workforce to 15,000 in a country where total tech employment is already tight requires either aggressive immigration or a training pipeline that produces results faster than the global talent market can absorb graduates. India's massive talent pool is one obvious source, but retaining those professionals against Silicon Valley salaries remains the challenge everywhere. The 2025 Oxford Insights index showed Singapore slipping to 7th place overall, suggesting that other nations are closing the gap on government AI readiness even as Singapore invests heavily.

And the governance-as-product model has a shelf life. AI Verify's value depends on it staying ahead of the technology it tests. The updated 2025 framework added generative AI considerations aligned with the 2024 Model AI Governance Framework for Generative AI, but the pace of model development means testing methodologies risk obsolescence within months of release. When new model architectures emerge faster than test suites can be written, the gap between what AI Verify can evaluate and what companies are actually deploying grows quietly. Japan's innovation-first regulatory approach offers a contrasting bet: skip the testing framework entirely and let the market sort it out.

What Other Countries Can Learn

Singapore's real innovation isn't any single program. It's the decision to treat AI governance as exportable infrastructure rather than domestic regulation. AI Verify doesn't tell companies what to do; it gives them tools to prove what they're doing. The Global AI Assurance Pilot doesn't impose Singapore's rules on other countries; it helps build shared testing norms that work across jurisdictions.

For small nations watching the AI arms race between the US, China, and Europe, Singapore offers a specific playbook: pick a layer of the AI stack where size doesn't matter, build practical tools rather than theoretical frameworks, open-source everything so adoption scales beyond your borders, and invest in the compute infrastructure that makes your governance claims credible. Whether Singapore can maintain that position as larger nations develop their own frameworks is the open question, but the model itself has already been adopted far beyond the city-state's borders.

Sources

Research Papers:

Industry / Case Studies:

Commentary:

Related Swarm Signal Coverage: