🌏 International Benchmarks
Singapore's AI strategy compared to leading economies.
📊 Overview Comparison
| Region | Core strategy | Year | Investment | Governance | Strength | AI ranking |
|---|---|---|---|---|---|---|
| 🇸🇬 Singapore | NAIS 2.0 | 2023 | S$2B+ government / US$26B+ tech giants | Framework + testing (AI Verify) | Governance-led, international hub | Tortoise #3, Oxford #2 |
| 🇭🇰 Hong Kong | Innovation & Technology Blueprint | 2022 | HK$20B+ | Voluntary guidelines, no dedicated law | Greater Bay Area bridge, 3000 PFLOPS supercomputing | — |
| 🇹🇼 Taiwan | AI Island Plan / AI Basic Act | 2025 | ~NT$100B (~US$3.1B) | Principles-based framework law (passed Dec 2025) | Semiconductor hegemon (TSMC) | — |
| 🇦🇪 UAE | AI Strategy 2031 | 2017/2021 | $100B MGX fund / $15.2B Microsoft | Voluntary ethics code, sandbox-friendly | Largest capital pool, world's first AI Minister | Tortoise #18, Oxford #3 |
| 🇮🇱 Israel | National AI Program | 2021 | NIS 5.26B (~$1.48B) but only 20% spent | Soft law + sector self-regulation, no horizontal legislation | Highest startup density globally, Unit 8200 talent pipeline | — |
| 🇰🇷 South Korea | K-AI Strategy / AI Basic Act | 2019/2025 | ₩100 trillion (~$71.5B) public-private fund | AI Basic Act (passed 2024) | Chaebols + semiconductors, dominant investment scale | Tortoise #7 |
| 🇪🇪 Estonia | Kratt AI Strategy | 2019 | €10M (extreme efficiency) | Pioneer in legal definition of AI Agents | World's #1 digital government, 50+ government AI use cases | — |
| 🇨🇭 Switzerland | Federal AI Strategy | 2020/2025 | CHF 1B+ research (ETH/EPFL) | Innovation-first, light-touch regulation | ETH/EPFL global Top 5, Google Zurich | Tortoise #9 |
| 🇫🇮 Finland | AI Finland / AuroraAI | 2017 | €100M+ AI business programme | Human-centric ethics, aligned with EU AI Act | Elements of AI national course, AuroraAI citizen services | — |
| 🇨🇦 Canada | Pan-Canadian AI Strategy | 2017/2024 | CAD $2.4B (2024 budget) | Voluntary code of conduct, AIDA bill shelved | Birthplace of deep learning; Mila/Vector/Amii | Tortoise #5 |
💡 Key Insights
Governance Models Diverge
Regions are visibly splitting between "legislation vs self-regulation". South Korea and Taiwan have opted for AI Basic Acts, the EU has taken a heavy-regulation route, while Singapore, Israel and Switzerland prefer flexible frameworks.
Investment Scales Vary Wildly
South Korea's ₩100 trillion and the UAE's US$100 billion MGX fund dwarf others. But Estonia proves €10 million can deliver 50+ government AI use cases — what matters is efficiency, not scale.
Talent Is the Core Variable
Israel's Unit 8200, Canada's Bengio/Hinton, Finland's mass AI education — every successful AI strategy is backed by a distinctive talent source.
Singapore's Distinctive Position
Singapore leads on governance maturity (AI Verify), execution discipline and international trust, but trails on investment scale, sovereign large models and fundamental research.
🔍 Region Profiles
🇭🇰 Hong Kong — Hong Kong Special Administrative Region of China
Hong Kong has committed over HK$20 billion to AI and innovation in recent years, including a 3000 PFLOPS supercomputing centre at Cyberport. But it lacks a unified AI strategy, with most major initiatives only launched in 2024-25 — a late-mover catch-up posture.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| AIRDI (AI R&D Institute) | HK$1 billion | Focused on applied R&D |
| Frontier Technology Fund | HK$3 billion | Supports frontier tech including AI |
| AI Subsidy Scheme | HK$3 billion | Subsidies for enterprise AI adoption |
| Innovation and Technology Fund | HK$10 billion | General-purpose tech fund |
⚖️ Governance Model
Hong Kong takes a voluntary-guidelines approach with no dedicated AI legislation. Regulatory authority is fragmented across the Digital Policy Office (DPO), the Privacy Commissioner for Personal Data (PCPD), the HKMA and other bodies, with no unified coordination. The common-law tradition provides some flexibility but also means rules are less explicit.
🎯 Key Initiatives
- Cyberport 3000 PFLOPS supercomputing centre
- AI Supercomputing Subsidy Scheme (AICP)
- Hong Kong AI R&D Institute (AIRDI)
- Smart Government Innovation Lab
- Fintech AI Sandbox (HKMA)
✅ Strengths vs Singapore
- • Greater Bay Area bridge — connecting the mainland's massive market with international capital
- • HQ of homegrown AI firms such as SenseTime
- • Common-law system; legal environment familiar to international firms
- • 3000 PFLOPS supercomputing plan exceeds Singapore's current compute
❌ Weaknesses vs Singapore
- • No unified national-level AI strategy
- • Fragmented regulation; agencies operate in silos
- • Late start; most key initiatives only launched in 2024-25
- • Geopolitical factors may affect international cooperation and talent flow
🏛️ Key Bodies
📚 Sources
- • Hong Kong Innovation and Technology Development Blueprint (2022)
- • AI-related policies in the 2024-25 Policy Address
- • PCPD Artificial Intelligence Ethical Framework (2024)
🇹🇼 Taiwan — Taiwan
Taiwan has put forward an "AI Island" vision, passed an AI Basic Act in late 2025, and committed over NT$100 billion in investment. As the undisputed global hegemon in semiconductor manufacturing (TSMC), Taiwan holds an irreplaceable strategic position in the AI hardware supply chain.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| AI Island Master Plan | NT$100 billion | About US$3.1 billion, multi-year investment |
| 2026 AI Budget | NT$30 billion | Annual government budget |
| AI Startup Programme | NT$10 billion | Support for startups |
⚖️ Governance Model
Taiwan passed the AI Basic Act in December 2025, taking a principles-based framework legislative approach, with the National Science and Technology Council (NSTC) as the competent authority. The Act emphasises innovation promotion, risk tiering, transparency and human rights, but detailed rules await secondary legislation.
🎯 Key Initiatives
- Continued TSMC advanced-node capacity expansion
- NCHC compute upgrade for AI workloads
- AI Basic Act legislation (Dec 2025)
- Ten Major AI Infrastructure Plan
- AI startup ecosystem cultivation
✅ Strengths vs Singapore
- • TSMC is irreplaceable in advanced AI chip manufacturing
- • Complete semiconductor and hardware ecosystem
- • Strong engineering talent pipeline
- • AI Basic Act provides a more explicit legal framework than Singapore's
❌ Weaknesses vs Singapore
- • Lacks globally significant AI software firms
- • Energy supply constrains compute expansion
- • Cross-strait geopolitical risk weighs on international confidence
- • Software and application layers are relatively weak
🏛️ Key Bodies
📚 Sources
- • AI Taiwan Action Plan 2.0 (2023)
- • AI Basic Act draft and Legislative Yuan records (2025)
- • Executive Yuan Ten Major AI Infrastructure Plan (2025)
🇦🇪 UAE — United Arab Emirates
The UAE was the first country in the world to appoint an AI Minister (2017), and has demonstrated formidable capital firepower through the US$100 billion MGX fund and a US$15.2 billion partnership with Microsoft. Falcon LLM and MBZUAI embody its ambition to build sovereign AI capability.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| MGX Fund | US$100 billion | Dedicated AI investment fund |
| Microsoft Partnership | US$15.2 billion | Cloud computing and AI infrastructure |
| Stargate UAE | 1GW data centre | Hyperscale compute project in partnership with the US |
⚖️ Governance Model
The UAE pursues a pro-innovation, light-touch regulatory path, relying primarily on non-binding ethical guidelines and regulatory sandboxes. It has the world's first AI Minister and a dedicated AI Office, but its regulatory framework is less mature than Singapore's AI Verify system.
🎯 Key Initiatives
- Falcon LLM open-source large model
- MBZUAI (Mohamed bin Zayed University of Artificial Intelligence)
- MGX US$100 billion AI investment fund
- Stargate UAE hyperscale data centre
- AI Minister role and the AI Office
✅ Strengths vs Singapore
- • Capital scale far exceeds Singapore — MGX US$100 billion vs Singapore government S$2 billion+
- • Cheap energy supports large-scale compute
- • Falcon LLM demonstrates sovereign large-model development capability
- • MBZUAI builds a world-class AI research university
❌ Weaknesses vs Singapore
- • Heavy reliance on foreign talent; weak local AI talent pool
- • Geopolitical sensitivity (chip export-control risk)
- • Regulatory framework still immature; lower international trust than Singapore
- • Academic research depth still trails Singapore's NUS/NTU
🏛️ Key Bodies
📚 Sources
- • UAE AI Strategy 2031 (2017, updated 2021)
- • MGX Fund official announcements (2024)
- • MBZUAI website and research reports
🇮🇱 Israel — State of Israel
The "Startup Nation" has the world's highest startup density in AI and the legendary Unit 8200 talent pipeline, but faces a severe execution gap — only 20% of the NIS 5.26 billion national programme has been spent, and there is no operational national supercomputer.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| National AI Program Budget | NIS 5.26 billion (~US$1.48 billion) | Five-year programme; only US$281 million actually spent (~20%) |
| Supercomputing Centre (Nebius) | US$140 million | Built jointly with Nebius; still under construction |
⚖️ Governance Model
Israel takes a soft-law approach favouring industry self-regulation, with no horizontal AI legislation. Sectoral regulators (banking, healthcare, etc.) issue their own AI guidance. Political instability has severely undermined the continuity and efficiency of policy execution.
🎯 Key Initiatives
- Unit 8200 AI talent incubation pipeline
- Five pillars of the National AI Program
- Nebius supercomputing centre construction
- Israel Innovation Authority (IIA) AI startup support
✅ Strengths vs Singapore
- • World's highest startup density; an extremely active AI startup scene
- • Military intelligence units such as Unit 8200 supply top-tier AI talent
- • Unicorns like Wiz (acquired by Google for US$32 billion) showcase entrepreneurial strength
- • Global leader in cybersecurity AI
❌ Weaknesses vs Singapore
- • National programme execution severely lags; only 20% of budget spent
- • No operational national supercomputer (Singapore has NSCC)
- • Political turmoil disrupts policy continuity
- • Small home market; firms typically list and scale overseas (especially in the US)
🏛️ Key Bodies
📚 Sources
- • Israel National AI Program (2021)
- • State Comptroller AI Report (2024)
- • AI Policy on Regulation & Ethics (2023)
🇰🇷 South Korea — Republic of Korea
South Korea is the most ambitious AI player among mid-sized economies — its ₩100 trillion (~US$71.5 billion) public-private fund vastly outstrips peer nations. Chaebols including Samsung, Naver and Kakao are actively developing proprietary large models, and the 2024 passage of the AI Basic Act signalled governance resolve.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| Public-Private Joint AI Fund | ₩100 trillion (~US$71.5 billion) | Multi-year public-private partnership fund |
| NVIDIA Partnership | US$3 billion | AI infrastructure and R&D collaboration |
⚖️ Governance Model
South Korea passed the AI Basic Act in 2024 (effective 2025) — Asia's first full-scope AI law. The Act uses a risk-tiered approach, sets up an AI Committee, requires impact assessments for high-risk AI, and tries to balance innovation. It is more legally binding than Singapore's voluntary framework.
🎯 Key Initiatives
- ₩100 trillion public-private AI fund
- AI Basic Act implementation (2025)
- Proprietary large models from Samsung / Naver / Kakao
- US$3 billion AI partnership with NVIDIA
- AI semiconductor sovereignty strategy
✅ Strengths vs Singapore
- • Crushing investment scale — ₩100 trillion is roughly 25x Singapore's government AI spend
- • Chaebol system enables rapid large-scale AI deployment (Samsung, LG, Hyundai, etc.)
- • Semiconductor manufacturing capability (Samsung, SK hynix)
- • AI Basic Act provides a stronger legal framework than Singapore's
❌ Weaknesses vs Singapore
- • Chaebol dominance may crowd out the startup ecosystem
- • Less internationalised than Singapore; weaker English-language environment
- • Less effective at attracting international talent and firms than Singapore
- • Population ageing poses long-term talent challenges
🏛️ Key Bodies
📚 Sources
- • K-AI Strategy (2019)
- • Full text of the AI Basic Act (2024)
- • Korea AI Semiconductor Strategy (2024)
🇪🇪 Estonia — Republic of Estonia
With an AI budget of just €10 million, Estonia has delivered 50+ government AI use cases — a paragon of extreme efficiency. As the global benchmark for digital government (99% of public services online), its Bürokratt virtual assistant and legal definition of AI Agents lead the world.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| AI Strategy Budget | €10 million | Minimal budget, maximal efficiency |
⚖️ Governance Model
Estonia is the first country in the world to provide a legal definition of AI Agents, permitting AI systems to perform specific government services as "digital assistants". As an EU member, it must align with the EU AI Act. Its governance model is known for pragmatism and technology-first execution.
🎯 Key Initiatives
- Bürokratt government virtual assistant
- 50+ government AI use cases deployed
- Pioneer of the AI Agent legal framework
- e-Residency digital identity system
- X-Road government data exchange platform
✅ Strengths vs Singapore
- • World's #1 digital government — 99% of public services online
- • Extreme efficiency: €10 million delivers 50+ AI use cases — a benchmark Singapore can learn from
- • Global leader in legal definition of AI Agents
- • Small-state agility — extremely short policy experimentation cycles
❌ Weaknesses vs Singapore
- • Tiny scale (1.3 million population); lessons may not transfer directly
- • No homegrown tech giants or major AI firms
- • R&D investment cannot match Singapore's A*STAR or AISG
- • Limited talent pool; reliant on EU talent mobility
🏛️ Key Bodies
📚 Sources
- • Estonia Kratt AI Strategy (2019)
- • e-Estonia official reports
- • Government AI Readiness Index
🇨🇭 Switzerland — Swiss Confederation
ETH Zurich and EPFL are global Top 5 AI research institutions, and Google Zurich is the company's largest European R&D centre. Switzerland is known for an "innovation-first, light-touch" stance and has not yet enacted standalone AI legislation, but leads the world in fundamental research quality.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| AI Research Funding (ETH/EPFL) | CHF 1 billion+ | Sustained funding via the federal institutes of technology system |
⚖️ Governance Model
Switzerland takes an innovation-first, light-touch path and has not enacted standalone AI legislation. The federal government prefers to govern AI through existing legal frameworks while closely tracking spillovers from the EU AI Act. Hosting major international bodies (WEF, ITU) makes it a key venue for global AI governance discussions.
🎯 Key Initiatives
- ETH AI Center
- EPFL AI research cluster
- Google Zurich (largest European R&D centre)
- Swiss AI Initiative
- WEF AI Governance Alliance (headquartered in Geneva)
✅ Strengths vs Singapore
- • ETH/EPFL — global Top 5 AI research institutions
- • Top-tier corporate labs including Google Zurich and Disney Research
- • International talent magnet — high salaries and quality of life
- • Influence over global governance through hosting of international bodies
❌ Weaknesses vs Singapore
- • AI startup scene less active than Singapore's (no Southeast Asian market hinterland)
- • Government less proactive in AI industrialisation than Singapore (e.g. AISG)
- • Federalism slows policy coordination
- • High costs may constrain large-scale AI infrastructure build-out
🏛️ Key Bodies
📚 Sources
- • Swiss Federal AI Strategy (2020/2025)
- • ETH Zurich AI Center Annual Report
- • OECD AI Policy Observatory — Switzerland
🇫🇮 Finland — Republic of Finland
Finland used the "Elements of AI" online course to train 1% of its population in AI literacy, pioneering global mass AI education. The AuroraAI citizen services platform embodies a "human-centric" vision for AI-powered government services.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| AI Business Programme | €100 million+ | Drives enterprise AI adoption |
⚖️ Governance Model
Finland adopts a human-centric, values-driven governance model, emphasising that AI should serve human welfare. As an EU member, it actively aligns with the EU AI Act. Its distinctive trait is embedding AI ethics into mass-literacy programmes rather than relying solely on regulation.
🎯 Key Initiatives
- Elements of AI mass-literacy course (reaching 1% of the population)
- AuroraAI citizen life-event services
- AI Business Finland enterprise transformation programme
- FCAI (Finnish Center for Artificial Intelligence)
✅ Strengths vs Singapore
- • Global pioneer in mass AI literacy — Elements of AI translated into 25+ languages
- • AuroraAI demonstrates an innovative model for AI public services
- • Highly digitalised social foundation (similar to Singapore)
- • Human-centric ethics orientation has built an international reputation for "responsible AI"
❌ Weaknesses vs Singapore
- • Small market (5.5 million population); limited AI industrialisation scale
- • No homegrown AI champion (Nokia diminished after pivoting)
- • EU AI Act compliance burden may constrain innovation flexibility
- • Winter climate and geography make it harder to attract Asian AI talent
🏛️ Key Bodies
📚 Sources
- • Finland AI Strategy (2017, updated 2019)
- • AuroraAI Programme Report
- • Elements of AI official statistics
🇨🇦 Canada — Canada
Canada is the birthplace of deep learning (Hinton, Bengio) and home to three world-class AI institutes — Mila, Vector Institute and Amii. The 2024 federal budget added CAD $2.4 billion in AI investment, but the AIDA legislation failed to pass, leaving governance reliant on voluntary codes.
📋 Core Strategies
💰 Investment
| Item | Amount | Note |
|---|---|---|
| 2024 Federal AI Budget | CAD $2.4 billion | Covers compute, safety, talent and commercialisation |
| Sovereign Compute Investment | CAD $1 billion | National-level AI compute infrastructure |
⚖️ Governance Model
Canadian AI governance relies primarily on voluntary codes; the proposed AIDA (Artificial Intelligence and Data Act) was shelved when Parliament was dissolved. Canada concentrates on frontier AI safety research through CAISI (Canadian AI Safety Institute) and plays a significant role in global AI safety governance.
🎯 Key Initiatives
- Mila (Montréal Institute for Learning Algorithms, led by Bengio)
- Vector Institute (Toronto, founded by Hinton)
- Amii (Alberta Machine Intelligence Institute)
- CAISI (Canadian AI Safety Institute)
- CAD $1 billion sovereign compute programme
✅ Strengths vs Singapore
- • Birthplace of deep learning — the academic legacy of Bengio and Hinton
- • Three world-class institutes form a talent development network
- • Global leader in AI safety and ethics research (CAISI)
- • World's first national AI strategy (2017); a clear first-mover advantage
❌ Weaknesses vs Singapore
- • Significant AI brain drain to the US (the "northbound brain drain" runs in reverse)
- • AIDA shelved; governance framework lacks legal force
- • Weak commercialisation — strong research, weak deployment
- • No homegrown AI giant (compare to Singapore's Grab or Sea)
🏛️ Key Bodies
📚 Sources
- • Pan-Canadian AI Strategy (2017/2024)
- • Budget 2024 — AI Chapter
- • CIFAR AI Strategy Reports
⚠️ Data on this page is compiled from official government documents, international organisation reports and public sources, independently curated by Singapore AI Observatory. Data as of February 2026.
Last updated: 2026-02-17