Taiwan
SI Index50.6/ 100
Indicative, not ranked
Data cover 7 of 8 pillars. The score is indicative and not ranked.
Experimental index, v1.1
Eight pillars
Compute 0: no cloud availability zone listed by OECD.AI. This does not mean that no AI compute is installed in the country.
How the coded scores were built
AI in schools (EDU-B)
| Component | Level (0–100) | Weight | Status |
|---|---|---|---|
| Curriculum | 50 | 35% | Evidence |
| Teacher training | 25 | 30% | Evidence |
| AI tools for students | 50 | 25% | Evidence |
| Monitoring | 50 | 10% | Evidence |
Curriculum 50: information technology compulsory (108 curriculum), explicit AI not for all (cap 50). Teachers 25: workshops and training, no national rate. Deployment 50: Ministry AI-assisted self-directed learning in 81 upper secondary schools. Monitoring 50: Academia Sinica study of grade 9 students.
Semiconductor sovereignty (SOV-B)
| Component | Points | Maximum | Status |
|---|---|---|---|
| Design capability | 12 | 25 | Evidence |
| Manufacturing leverage | 25 | 25 | Evidence |
| Resilience of access | 20 | 20 | Evidence |
Design 12: MediaTek, Alchip and GUC design AI accelerators produced in series (including for hyperscalers). Manufacturing 25 (cap): TSMC leading-edge foundry (A3) + CoWoS packaging (A5) + ASE assembly (B7). Resilience 20: free access (14) + production on the territory (6).
Economic outcomes
Not part of the score
- Labour productivity
- $69.9 per hour worked (2025)+4.0% a year since 2019
- GDP per person
- Not available
Productivity in constant 2021 international dollars at purchasing power parity. Sources: ILO modelled estimates (Nov. 2025); World Bank WDI.
Education pillar
Based on PISA and AI-in-schools coding
Detailed indicators
Scores are normalised to 0–100 against fixed bounds. Raw values are shown as published by each source.
| Indicator | Score (0–100) | Raw value |
|---|---|---|
| Compute | ||
| CMP-B AI compute commercially available in the cloud | 0.0 | 0 Cloud availability zones, weighted by tier |
| Energy | ||
| NRG-A Electricity generation per person | 87.3 | 12,400 kWh per person |
| NRG-B Growth in generating capacity | 45.0 | 18.9 Capacity growth over five years (%)0.429 GW added per million people |
| Talent | ||
| TAL-B Highly cited AI research per million people | 83.7 | 58 Top-10% cited AI publications per million people |
| Education | ||
| EDU-A Top performers in PISA 2025 | 58.3 | 35 Share of top performers (%) |
| EDU-B AI in schools (coded) | 42.5 | — |
| Capital | ||
| CAP-A Private AI investment | Not available | — |
| CAP-B Depth of capital markets and private credit | Not available | — |
| Adoption | ||
| ADO-A Working-age population using AI | 33.8 | 33.8 Share using AI (%) |
| Institutions | ||
| INS-A Regulatory quality | 76.1 | 76.1 WGI 2025 score |
| INS-B Government effectiveness | 83.3 | 83.3 WGI 2025 score |
| Sovereignty | ||
| SOV-A Notable AI models developed | 0.0 | 0 Models, last three years (Epoch AI) |
| SOV-B Semiconductor sovereignty (coded) | 81.4 | — |