🎯 National AI Levers

Last updated: 2026-05-03. Re-classifying Singapore's Budget 2026 plus all ministry-level AI policies along "AI-injection paths" yields six cross-ministry levers — a clearer view of the whole shape than reading them department by department.

Why classify by lever?

Most analyses treat the state and the enterprise as parallel actors — and miss the nesting relationship. The essence of Singapore's strategy is to treat the entire state as the wrapper layer for enterprise AI-native transformation: the state itself need not become AI-native; it is enough to amplify the speed at which enterprises do.

The six levers map "AI-injection paths" (infrastructure → governance → talent → applications → government self-use → diplomacy). Each lever crosses multiple ministries; chained together, they form the full execution pipeline.

🏗️

Lever 1 · Infrastructure

Transmitted to enterprise

Data + Compute + Physical Infrastructure

What the state does
Build the compute and data layer companies cannot afford to build themselves
Enterprise bottleneck solved
Companies cannot afford their own compute

Foreign-Capital Compute (EDB)

S$5.5B

2024–2028 investment agreement, covering AI skilling and local infrastructure

S$12B

2024–2028 investment, expanding Singapore cloud infrastructure capacity

US$9B / ~S$11.6B

Includes a DeepMind lab on the ground

Multiple partnerships

NVIDIA has built a multi-institution AI acceleration network in Singapore

Established

APAC HQ based in Singapore, deepening collaboration with AISG

Locally Subsidized Compute

S$150M

Direct subsidies for enterprises to purchase AI compute

Physical infrastructure

Launched in Budget 2026, an AI company cluster zone

Capital Platforms

S$1.5B

Launched in 2026, an IPO anchor fund supporting local tech firm listings

S$1.5B

Launched in 2026, a development capital pool for emerging industries

S$6.5B

Expanded allocation in 2026, drawing private capital into local tech investment

National Data Substrate

HEALIX

MOH / Synapxe
National health data + AI infrastructure

A national substrate that makes medical records consumable by AI

National digital twin

A digital mirror for urban planning and operations, paired with ePlanner 3D + Smart Planning Assistant

First fully smart district

Expected 30% energy reduction; a national-level smart district template

Homeland Security Compute

Owned compute

MHA's tech agency builds its own supercomputer, supporting police / immigration / security AI

💡 The state turns the biggest fixed costs (compute + data) into public goods.
⚖️

Lever 2 · Governance

Transmitted to enterprise

Rules + Sandboxes + Law

What the state does
Give companies enough certainty to deploy
Enterprise bottleneck solved
Companies hold back because compliance risk is unmeasurable

General Governance Frameworks (IMDA)

2019

Asia's first AI governance framework

10+ multinational enterprises participating

The world's first AI governance testing framework and toolkit

2024–2025

Generative AI evaluation and experimentation environment

2024

Nine dimensions; one of the earliest dedicated GenAI governance frameworks globally

Released at Davos on 2026-01-22

The world's first Agentic AI governance framework

SS 714:2025

Trusted data sharing framework, with national standard upgraded

Financial Sector Governance 5-Layer Stack (MAS)

2018

Four principles: Fairness / Ethics / Accountability / Transparency

2021–

Open-source toolkit for FEAT principles, pushing principles toward operability

24 institutions + Microsoft / AWS / Google / NVIDIA

Co-built GenAI risk framework for the financial industry

2024-12

Formal supervisory expectations document constraining bank use of AI

Successor platform

Lets regulated financial institutions continuously test and report AI risk

Cybersecurity Governance (CSA)

2024-10

Full-lifecycle security guidelines for AI systems

2025 (public consultation)

Security addendum focused on Agentic AI

2025

Frontier AI risk advisory, coordinated with AISI

AI dimension extension

Cybersecurity certification extended to AI

Legal Governance (MINLAW + IPOS)

2021

Tied with Japan as the world's most permissive copyright stance on AI training

2024

Authorship stance on AI-generated content

2024–2025

OCHA + Elections Bill 2024 (deepfake ban) + Criminal Law Bill 2025 (criminalizing AI-generated intimate imagery) + Online Safety (Relief and Accountability) Bill 2025

💡 "Permissive on training, strict on outputs" — Japan and Singapore are the only two countries doing this.
🎓

Lever 3 · Talent

Transmitted to enterprise

Education + Training + Transition

What the state does
Make sure companies can find people who can actually use AI
Enterprise bottleneck solved
Companies cannot find people who can use AI

Whole-Population Layer (MDDI)

100K workers; first cohort 1H 2026

First cohorts: Accounting (ISCA) + Legal (SAL/SCCA)

Population-wide AI literacy

Launched in Budget 2026, covering primary and secondary students through seniors

Professional Layer (IMDA + AISG)

21K placed + 340K upskilled

Tilts the national skills accelerator toward AI

16 cohorts, 410+ apprentices; new cohort with 800 slots

Industrial-grade AI engineer training

S$150K co-investment per project, 100+ completed cumulatively

Enterprise × AISG joint AI projects

Education System Layer (MOE + NIE)

8 categories of tools

AI integration for the national K-12 learning platform

2024–2026

GenAI usage rules at the school level

Released in 2024

Writes AI into the next 5 years of education technology planning

Teacher AI literacy certificate

Launches AI teaching capacity-building from the teacher side

Fully rolled out

AI curriculum upgrade across all levels from undergrad to EMBA

Fiscal Subsidy Layer (SSG + WSG)

Tiered at 50% / 70%

105K+ enrolled in 1,600+ AI courses (2025)

S$4,000

Dedicated subsidy for mid-career retraining

2025

One-stop skills and employment platform

Mid-Career Retraining Layer (MOM)

Job Redesign+

MOM / WSG
"Human + AI job redesign"

Policy framework that converts "layoffs" into "job redesign"

Mid-career retraining

Helps PMETs transition into AI-related roles

Enterprise level

Enterprise-level workforce AI transformation support

💡 Whole population, enterprises, schools, mid-career retraining — the talent pool only works when all five layers line up.
🚀

Lever 4 · Applications

Transmitted to enterprise

Industry + Public Service Deployment

What the state does
Roll out flagship applications simultaneously across 11 ministries
Enterprise bottleneck solved
Enterprises unwilling to move + insufficient public-service demonstration

Industry Flagships (MTI)

4 pioneer industries

Focused on industry-level AI application breakthroughs

Multi-institution consortium

AI centres of excellence linking industry, academia and research

Embodied AI R&D

MTI / A*STAR
R&D programme

R&D fusing robotics + AI

Research Flagships (A*STAR)

5 research pillars

Frontier AI research centre

Mencast propeller case study

AI application deployment in manufacturing

50–100x acceleration

AI-accelerated materials science research

S$70M

National-level multimodal large-model research

Regional LLM Flagships (AISG)

11+ languages, 4B–33B parameters

Southeast Asian multilingual foundation model

SEALD

AISG
Regional language dataset

Southeast Asian AI training dataset

Enterprise Diffusion (IMDA + ESG)

10K enterprises + 100K workers / 2026–2029

National-level enterprise AI diffusion programme

Champions of AI

IMDA + ESG
Flagship enterprises

AI transformation programme for flagship industry leaders

30% → 50%

Productivity Solutions Grant AI category subsidy rate raised

SMEs

AI enablement package for SMEs

Healthcare (MOH + Synapxe)

5,000+ clinicians, 67K notes (2025-12)

Generative AI helps doctors write clinical notes

4.5/5 rating (public beta)

Consumer-facing AI health assistant

AimSG

Synapxe
National medical imaging AI

National medical imaging AI platform

SELENA+

Synapxe
Nationwide deployment

AI for diabetic retinopathy screening

Rollout to 1,100+ Healthier SG clinics in early 2027

3-year risk prediction for diabetes / hyperlipidemia

National coronary AI platform

Cardiovascular imaging AI

Transport (MOT + LTA + PSA + CAG)

3 routes, launched 2025-12

First batch of commercialized autonomous public transit

National-grade test facility

one-North autonomous-driving test area

World's largest fully automated port by the 2040s

A port built on AI + automation + predictive analytics

World-first

Benchmark for airport AI governance certification

Construction & City (MND + HDB + BCA + URA + JTC)

S$30M

Centre of Excellence for AI in construction

National digital delivery standard

End-to-end digitalization from design to operations & maintenance

42,000 households

First smart energy town

Virtual assistant

AI assistant for resident services

Environment & Water (MSE + NEA + PUB)

S$25M

AI-augmented extreme weather forecasting

Nationwide deployment

AI predicts high-risk outbreak zones

Nationwide rollout

Smart water meters + AI leak detection

Real-time flood monitoring

Nationwide real-time drainage monitoring centre

💡 Flagship applications rolling out across 11 ministries at once — this is what a national-scale "queryable organisation" looks like in practice.
🏛️

Procurement / Leading by Example

What the state does
Put civil servants on AI first, so businesses have a precedent
Enterprise bottleneck solved
Businesses have no precedent to point to

Civil Government (GovTech)

Target: 150K civil servants

AI work assistant for the entire civil service

Hansard + Courts + Legislation

Makes all public government artifacts AI-readable

LaunchPad

GovTech
3K MAU / 400 ideas

AI experimentation platform for civil servants

Ministry-level AI deployment projects

Scaling successful GenAI use cases from pilot to production

AI safety dual-pack

Safety and protection platform for government AI deployments

First in Asia

Agentic AI inside an air-gapped environment

Defence (MINDEF + DSTA + DSO + DIS)

2022 fourth service branch

Reorganized in 2025 into DCCOM + SAFC4DC — writing AI into the structure of the service branch itself

Military–national research collaboration

Military AI curriculum and project collaboration

Multiple commercial AI partnerships

Autonomous drones, AI Co-Lab, Lattice manned-unmanned teaming

Naval AI

AI-assisted naval ship identification

In-house + top-tier collaboration

Defence GenAI toolchain

Partnership with Mistral

GenAI projects for defence scenarios

Homeland Security (HTX + SPF + ICA)

Self-trained

MHA's in-house LLM

S$100M (launching Q2 2026)

National R&D centre for humanoid robotics

Multiple commercial partnerships

Plugged into the commercial cloud + AI ecosystem

RPA + AI

Scam detection and automated response

City surveillance + airport robots + smart glasses

Real-time video analysis and automated response

Iris + face

Multi-modal biometrics at the border

💡 DIS writes AI into a service branch’s structure and GovTech Pair turns every civil servant into an AI operator — the IC + DRI + AI Founder model, at national scale.
🌐

International Governance + Foreign Capital + Standard-Setting

What the state does
Get foreign firms to base their AI governance HQ in Singapore
Enterprise bottleneck solved
Companies don't know where to place their HQ

Global Frameworks Initiated by Singapore

35+ national bodies, 250+ experts (17th SC 42 plenary, Apr 2026)

World's first proposed international standard for testing generative AI; benchmarking + red teaming

S$10M/year

Frontier AI safety research + coordinating centre for the Singapore Consensus

11 countries signed (including US and China)

Rare in bringing both the US and China into a single AI safety document

2024 + 2026

Global AI safety scientific exchange convened alongside ICLR

Global red-teaming challenge

Standardizes global red-teaming evaluation methods

ASEAN Region

Regional working group

Singapore serves as secretariat

ASEAN Guide on AI Governance and Ethics

ASEAN Digital Ministers / IMDA
Adopted by 10 countries

Based on Singapore's Model AI Governance Framework

ASEAN Hanoi Declaration 2026

ASEAN Digital Ministers
2026

Deepening cooperation among regional digital ministers

Bilateral Cooperation

Bilateral cooperation framework

Smart Cities Programme + Digital Economic Cooperation Roadmap

Cooperation with South Korea

Bilateral agreement on AI technology and governance

EU × ASEAN

Inter-regional AI governance dialogue mechanism

Military / Security

Co-hosted by Singapore

Extending the REAIM dialogue to Asia

Co-hosted by 5 countries

"Blueprint for Action" — the first operational multilateral document on military AI governance

Participated in all

Sustained voice across all three AI safety summits

United Nations + Global

Independent International Scientific Panel

UN-level AI governance dialogue and scientific panel

Global programme

Exporting Singapore's AI education model to developing countries

💡 Punching at G7-level AI governance influence with 0.07% of the world's population — the least replicable part of Singapore's strategy.

Key Observation: 6 Levers → 7 Transmission Channels

Re-slicing the 6 levers by "which enterprise bottleneck they solve" yields 7 transmission channels:

1. Pull (capital returns) → Lever 1 (ECI, PSG) + Lever 2 (sandboxes that make risk measurable)
2. Push (forward pressure) → Lever 4 (NAIIP 10K + Champions of AI)
3. Talent (talent pool) → Lever 3 (AI Bilingual 100K + AIAP + university curriculum reform)
4. Infra (compute substrate) → Lever 1 (EDB hyperscaler attraction + ECI + one-north)
5. Trust (deployment boundary) → Lever 2 (IMDA + MAS + CSA + MINLAW)
6. Procurement (leading by example) → Lever 5 (GovTech + DIS + HTX)
7. International (foreign capital + governance HQ) → Lever 6 (AISI + Singapore Consensus + ASEAN)

Of the 7, only #6 (Government Self-Use) and #7 (Diplomacy) are done directly by the state — the other 5 are levers the state uses to transmit through to enterprises.

This is the structural evidence for the thesis: "the state cannot be AI-native on its own; it must transmit through to enterprises."