A policy analysis: July 2026
Singapore has no domestic tech giant on the scale of a Google or an Alibaba, no chip fabs comparable to Taiwan’s, and a population smaller than metropolitan Houston. What it has instead is a habit of treating national strategy as a living document rather than a one-time announcement, and a willingness to fund that document at a scale most city-states never attempt. In December 2023, Prime Minister Lawrence Wong launched the National AI Strategy 2.0, the country’s second attempt at a comprehensive AI roadmap since its first strategy in 2019. In May 2026, barely two and a half years later, Minister for Digital Development and Information Josephine Teo returned to the same ATxSummit stage to announce a refresh, describing it not as a system reboot but as a double-click on what was already working. That cadence, launch, learn, and relaunch, is the closest thing Singapore has to a signature move in AI policy.
From NAIS 2.0 to the 2026 Update
NAIS 2.0 was built around a single guiding vision, “AI for the Public Good, for Singapore and the World,” organised into two central goals: Excellence, meaning Singapore would selectively build peaks of capability rather than try to compete everywhere at once, and Empowerment, meaning individuals and businesses across the economy would be equipped to use AI rather than watch it happen elsewhere. The strategy translated those goals into three Systems, Industry, Government, and Research, supported by ten Enablers spanning Talent, Compute, Data, Trusted AI, and international partnerships, and fifteen concrete Actions to be delivered over three to five years. Five sectors were named as early priorities for adoption: healthcare, smart cities and estates, education, safety and security, and logistics.
By early 2026, Singapore judged that the pace of global AI development had outrun even that framework. In February 2026, the government established a National AI Council, chaired personally by PM Wong, to give the strategy sharper political direction. Three months later, at ATxSummit 2026, Teo unveiled the Update to NAIS, setting out ten refreshed priorities across industry, government, research, talent, workforce capability, ecosystem integration, compute, data, trust, and international partnerships. The update kept the original vision statement intact. What changed was the emphasis, shifting the strategy’s centre of gravity from building foundational capacity toward converting that capacity into sector-wide, production-grade deployment.
The headline mechanism for that shift is a new generation of National AI Missions, targeting Advanced Manufacturing, Financial Services, Connectivity, and Healthcare, four sectors that together account for over 40 percent of Singapore’s GDP. Rather than funding scattered pilots, the missions are designed to concentrate government, industry, and research resources on a small number of sectors where success can be measured and reproduced elsewhere in the economy.
The Numbers Behind the Pivot to Adoption
Singapore’s decision to pivot toward deployment was backed by adoption data that told two different stories depending on company size. The Singapore Digital Economy Report, published by the Infocomm Media Development Authority (IMDA) in October 2025, found that AI adoption among small and medium-sized enterprises had reached 14.5 percent in 2024, up from 4.2 percent the year before, a more than threefold increase. Among larger, non-SME companies, adoption jumped from 44 percent to 62.5 percent over the same period. The gap between big firms and small ones remained wide even as both curves moved upward quickly.
That gap is the direct rationale for the National AI Impact Programme (NAIIP), announced by MDDI at the Committee of Supply Debates in March 2026. NAIIP commits to supporting 10,000 enterprises over three years to advance their AI adoption, paired with a workforce-side target of training 100,000 workers to become “AI Bilingual,” a term Singapore uses to describe workers who combine domain expertise with practical fluency in applying AI to their own workflows. The programme layers several delivery mechanisms on top of that headline target. IMDA is enhancing its existing Digital Leaders Programme with a new Digital Leaders Accelerator Bootcamp, aimed at building the confidence of business leaders through hands-on AI project work rather than slide-deck briefings. Working with Enterprise Singapore, IMDA has pre-approved a curated list of proven, cost-effective AI solutions eligible for Productivity Solutions Grant support, and plans to raise the share of AI-enabled solutions on that list from 30 percent to 50 percent. On the talent side, the TechSkills Accelerator programme, which has placed more than 24,300 people into tech roles and upskilled over 440,000 individuals since its 2016 launch, is being extended to build AI fluency among non-tech professionals, starting deliberately with accountancy and legal roles chosen because they cut horizontally across nearly every industry and carry high exposure to AI-assisted work. IMDA is partnering directly with the Institute of Singapore Chartered Accountants, the Singapore Academy of Law, and the Singapore Corporate Counsel Association to design the specific curricula.
Research: SEA-LION, MERaLiON, and a Billion-Dollar Bet on Fundamentals
Underneath the adoption push sits a research strategy that predates NAIS 2.0 by six years. AI Singapore (AISG), launched in 2017 by the National Research Foundation with an initial five-year commitment of up to S$150 million, has since been extended through 2027 and become the connective tissue between Singapore’s universities, research institutes, and AI start-up ecosystem. Its best-known output is SEA-LION, Southeast Asian Languages in One Network, a family of open-source large language models purpose-built for the region’s linguistic diversity, developed in part with Google DeepMind’s Gemma models as a technical foundation. SEA-LION has been downloaded millions of times and, according to AISG’s own benchmarking, has outperformed larger general-purpose models on regional language tasks since its earliest releases. A parallel effort, MERaLiON, led by the A*STAR Institute for Infocomm Research under Singapore’s S$70 million National Multimodal Large Language Model Programme, focuses on multimodal, empathy-aware speech and language processing tuned to Singaporean and regional accents; it has been downloaded more than 90,000 times since its December 2024 launch and is already being applied to elder-care assistants and scam-call detection.
In January 2026, at Singapore AI Research Week, the government updated its National AI R&D (NAIRD) Plan and, with support from the National Research Foundation, committed more than S$1 billion to public AI research and talent development between 2025 and 2030. That figure spans fundamental research, applied research, and the talent pipeline feeding both, and it is explicitly framed in the NAIS update as the resourcing needed to sustain an AI hub rather than merely an AI-adopting economy. Alongside AISG, an eminent international advisory group, including Yoshua Bengio, Stuart Russell, Dawn Song, Zhang Ya-Qin, and Max Tegmark, updated the so-called Singapore Consensus on AI Safety Research Priorities this year, giving the country’s safety research agenda a level of global academic input that few nations of its size can claim.
Betting the House on Foreign Capital and Compute
Singapore’s AI hub strategy has always depended on convincing the world’s AI leaders to build in Singapore rather than merely sell into it. That approach reached a new scale in May 2026. At ATxSummit, OpenAI and MDDI announced a partnership worth more than S$300 million, including OpenAI’s first Applied AI Lab outside the United States and a commitment to create more than 200 Singapore-based technical roles, positioning the city-state as one of OpenAI’s global hubs for forward-deployed engineering talent. Google signed a parallel memorandum of understanding with MDDI to co-develop AI solutions across the public and private sectors. Both deals followed a now-familiar pattern in Singapore’s playbook: rather than only funding domestic champions, the government actively recruits frontier labs to embed technical operations, and often applied research, inside its own borders.
Compute access for the rest of the economy runs through the Enterprise Compute Initiative, a S$150 million programme that helps companies secure cloud compute, tooling, training, and engineering support from a panel of cloud service partners, aimed squarely at firms that cannot negotiate GPU access on their own. Physically, Singapore is also building out dedicated testbeds: the Punggol Digital District is being developed as a frontier site for embodied and applied AI, while JTC’s planned Kampong AI campus at LaunchPad @ one-north, piloting from March 2026 with completion targeted for 2028, will offer roughly 14,500 square metres of space for up to 70 AI start-ups and accelerators.
Governance as a Selling Point, Not a Constraint
Where the European Union has built its AI credibility around binding legislation, Singapore has built its around voluntary, exportable tooling. AI Verify, a testing framework that benchmarks AI systems against eleven internationally recognised governance principles, has been available since 2022 and was updated to cover generative AI applications as the technology moved from narrow classifiers to general-purpose models. Project Moonshot, an open-source large language model evaluation toolkit combining benchmarking with red-teaming, is positioned by its developers as one of the first tools of its kind anywhere, giving compliance teams and model developers a shared way to stress-test systems before deployment. Both tools are designed for adoption well beyond Singapore’s borders, reflecting a governance philosophy that treats trust infrastructure as an export product rather than a domestic compliance cost.
Internationally, Singapore has leaned into the same pathfinder logic. Under the US-Singapore Critical and Emerging Technology Dialogue, the two countries are deepening information-sharing on AI security, safety, and standards, and Singapore’s official language explicitly favours starting small with like-minded bilateral partners as a route toward broader multilateral cooperation, rather than waiting for a global consensus to form first.
Over 70 Centres of Excellence, and the Test Still Ahead
Since NAIS 2.0 launched, more than 70 companies across sectors from advanced manufacturing to financial services have established AI Centres of Excellence in Singapore, dedicated teams focused on AI development and deployment in support of business goals; Nvidia and KPMG are among the more recent firms to join that list. Those centres, together with the OpenAI and Google commitments, the SEA-LION and MERaLiON research programmes, and the more than S$1 billion committed to AI R&D through 2030, give Singapore a plausible claim to being Southeast Asia’s default AI hub. What the country has been more candid about is the harder problem sitting underneath the headline numbers: turning SME adoption rates that, even after tripling, remain well under half the level seen in large enterprises, and converting a wave of pilot projects into the kind of durable, sector-wide deployment the new National AI Missions are explicitly designed to force. Singapore’s own framing of its 2026 update, a double-click rather than a reboot, is itself the tell. The country is not repositioning because its original strategy failed. It is repositioning because, having built the foundations faster than most nations its size, it has run out of foundation-building left to do and now has to prove the harder thing: that all of it adds up to an economy genuinely transformed, not just an ecosystem well resourced.
Key References and Further Reading
