Singapore's National AI Strategy and Governance

Singapore's National AI Strategy (NAIS) is the Government's policy framework for using and developing artificial intelligence for public benefit, economic activity and trusted adoption. NAIS 2.0 was launched in December 2023; the Smart Nation page says a 2026 update refreshed ten priorities after the AI landscape changed, and that the National AI Council was established in February 2026 to provide strategic direction for the national AI agenda (Smart Nation Singapore, updated 21 May 2026). The strategy is therefore a policy programme, not a single statutory regulator or an automatic entitlement for businesses.

What the strategy covers

NAIS 2.0 organised Singapore's effort around AI activity drivers, people and communities, and infrastructure and environment. In practical terms this links research and talent to business and government adoption, compute and data infrastructure, international partnerships, and responsible-use safeguards (MDDI, 4 Dec 2023; Smart Nation Singapore, updated 21 May 2026). The 2026 update also highlights industry AI Centres of Excellence, broad-based government adoption, AI research activity, AI governance tools and access to compute; the Smart Nation page reported more than 70 companies with AI Centres of Excellence in Singapore at the time of that update (Smart Nation Singapore, updated 21 May 2026).

Research and talent: the NAIRD Plan

The National AI Research and Development (NAIRD) Plan adds more than S$1 billion over five years from 2025 to 2030 for Singapore's public AI research capabilities. MDDI describes three focus areas: fundamental AI research, applied AI research, and talent development from pre-university level through faculty and international research partnerships (MDDI, 24 Jan 2026). This plan sits within the wider Research, Innovation and Enterprise framework and is distinct from enterprise grants, SkillsFuture training, or a general promise that every AI project will receive funding.

The investment is aimed at strengthening public research capability and Singapore's position as an AI research hub; it is not a single application window described in the announcement. Its stated research areas include responsible AI, resource-efficient AI, emerging AI methods and general-purpose AI, with applied work intended to support industry and RIE domains (MDDI, 24 Jan 2026).

Embodied and efficient AI research context

At ATxSummit 2026, MDDI said NVIDIA’s new Singapore research lab would focus on embodied AI and efficient AI, working with universities, industry partners and government agencies. The speech placed this work in the context of advanced manufacturing, where physical AI, simulations and digital twins may support process redesign, predictive maintenance and reduced material waste; these are collaboration and research directions, not a published grant entitlement or regulatory approval for every AI project (MDDI, 20 May 2026).

“Embodied AI” in this context refers to AI used with physical systems, while “efficient AI” concerns making advanced capabilities more usable under practical resource constraints. The speech identifies areas for research and collaboration; it does not establish that every manufacturer can access NVIDIA's lab, that a digital twin is required for an industrial project, or that a research collaboration has already produced a deployable product. A company still needs the relevant technical, safety, data and sector approvals for its own use case (MDDI, 20 May 2026).

Enterprise and workforce adoption

The National AI Impact Programme (NAIIP) is an enterprise-and-workforce adoption measure under NAIS 2.0. MDDI's 2 March 2026 factsheet sets a programme target of supporting 10,000 enterprises over three years to advance their AI adoption journey and 100,000 workers to become “AI Bilingual” — able to deepen AI fluency and apply AI in their domains (MDDI, 2 March 2026). These are programme targets, not a promise that every enterprise receives a grant or that every worker receives the same training benefit.

NAIIP combines several support mechanisms rather than one universal application product. MDDI says IMDA will launch a Digital Leaders Accelerator Bootcamp for business and technical confidence, expand the range of IMDA pre-approved AI solutions available with Productivity Solutions Grant support in partnership with EnterpriseSG, and enhance TechSkills Accelerator (TeSA) to build domain-specific AI capability. Initial professional focus includes accountancy and legal work, while a separate AI-fluency programme is planned for the tech workforce (MDDI, 2 March 2026). Current eligibility, training partners, application routes and course durations must be checked with the implementing programme because the factsheet says further details will be released separately.

NAIIP should not be collapsed into the National AI Research and Development Plan, which funds public AI research and talent; National AI Missions, which organise sector-level problem-solving ecosystems; or Champions of AI, which works with a selected group of leading companies on enterprise-wide transformation. NAIIP is the broad adoption-and-workforce layer: its targets describe reach across enterprises and workers, while the individual support mechanisms have their own conditions and administrators.

This distinction matters for practical questions. A business cannot infer a grant, procurement award, tax treatment or regulatory clearance merely from the existence of NAIS. The relevant agency or programme page must establish the current target group, application route, qualifying activity and applicable period before a company relies on an announcement.

Champions of AI: enterprise-wide transformation

The Champions of AI Programme is a national enterprise-transformation initiative jointly led by Enterprise Singapore (EnterpriseSG) and Digital Industry Singapore (DISG). It is intended for a select group of leading Singapore-based companies whose leadership is prepared to make AI a core driver of productivity, innovation and revenue growth. The programme goes beyond a small pilot: participating companies are expected to work toward enterprise-wide transformation across business functions, systems and workforce practices (Enterprise Singapore, Champions of AI Programme, accessed 19 Aug 2026; Enterprise Singapore, 2 Mar 2026).

Support is customised rather than an automatically claimable grant. EnterpriseSG describes long-term transformation roadmaps, implementation of priority AI projects, leadership and workforce training, internal AI-team capability, change management, and curated partnerships with experts or solution providers. Companies should have modern digital and data infrastructure and be ready to commit sustained resources to organisational change and upskilling. EnterpriseSG and DISG said they would pilot with a few companies first and expand over time, with shortlisted companies notified after registering interest (Enterprise Singapore, accessed 19 Aug 2026). Champions of AI therefore differs from the National AI Missions: Champions works with selected companies on organisation-wide transformation, while Missions organise sector ecosystems around shared problem statements.

National AI Missions

National AI Missions are the 2026 implementation model for applying AI to sector-level problems in four priority areas: advanced manufacturing, financial services, connectivity, and healthcare. MDDI describes them as part of the refreshed National AI Strategy, with missions organised around concrete problem statements and tangible outcomes rather than a collection of disconnected technology pilots (MDDI, 20 May 2026).

The model is intended to assemble sector ecosystems around the problem: relevant firms and public agencies, research and engineering capabilities, data, compute, and where needed regulatory sandboxes. MTI describes the related AI Park or “Kampong AI” concept as a way to bring these capabilities together so that solutions can move from research and experimentation toward adoption in Singapore's economy (MTI, Committee of Supply Debate 2026). A National AI Mission announcement is therefore a policy and coordination framework, not by itself a grant, automatic procurement award, or approval for every AI product; the responsible agency must publish the conditions for any specific call, project or support measure.

Governance is guidance and accountability, not one AI Act

Singapore's AI governance approach uses practical frameworks, testing tools and sector or data-protection rules rather than treating NAIS as a single licensing statute. IMDA's Model AI Governance Framework for Agentic AI, launched on 22 January 2026, guides organisations deploying AI agents that can plan across multiple steps. It recommends bounding autonomy and tool or data access, defining meaningful human-approval checkpoints, applying technical controls throughout the agent lifecycle, and supporting transparency and user education; organisations remain accountable for deployment decisions (IMDA, 22 Jan 2026).

The framework is guidance for responsible deployment, not proof that an AI system is approved by the Singapore Government. AI Verify and other testing resources can help organisations assess responsible implementation, while the Personal Data Protection Act, sector rules, cybersecurity requirements, consumer law and employment law may apply separately depending on the use case (Smart Nation Singapore, updated 21 May 2026; IMDA, 22 Jan 2026).

How to interpret Singapore AI announcements

Keep four layers separate: NAIS is the national direction; NAIRD is a public research-and-talent investment plan; the National AI Impact Programme is an adoption measure; and IMDA frameworks are governance guidance and tools. A company, researcher or worker therefore needs the current programme administrator and its published conditions for an actionable answer. Broader context on startups and research is in Singapore's startup ecosystem and A*STAR and public research, while data-centre capacity is an infrastructure issue covered in the data-centre industry.

These layers can move on different timelines: the strategy may be refreshed, a research plan may cover a stated multi-year period, an adoption programme may publish new guidance, and a governance framework may be revised without changing the underlying strategy. Answers should therefore name the layer, date and issuing institution instead of treating every AI announcement as the same policy instrument (MDDI, Update to Singapore’s National AI Strategy factsheet, 20 May 2026; MDDI, National AI Research and Development Plan, 24 January 2026; IMDA, Model AI Governance Framework for Agentic AI, 22 January 2026).