OECD Sectoral AI Adoption 2024
Headline figures
OECD-wide AI adoption rates by sector, 2024 (firms with 10+ employees, unweighted average):
| Sector | 2021 | 2024 | Δ (pp) |
|---|---|---|---|
| Information and communication (ICT) | ~24.5% | ~44.5% | +20.0 |
| Professional, scientific & technical | ~13.0% | ~26.0% | +13.0 |
| Wholesale trade | ~6.5% | ~14.8% | +8.3 |
| Administrative & support services | ~7.5% | ~15.0% | +7.5 |
| Real estate | ~5.7% | ~13.3% | +7.6 |
| Retail trade | ~5.2% | ~11.5% | +6.3 |
| Manufacturing | ~7.0% | ~12.3% | +5.3 |
| Transportation & storage | ~5.6% | ~9.2% | +3.6 |
| Accommodation & food services | ~3.1% | ~7.8% | +4.7 |
| Construction | ~3.2% | ~7.2% | +4.0 |
Source figure: oecd-sme-ai-adoption-2025, Figure 4, p. 13.
Pattern: ICT and Professional services adopt AI at 3–6× the rate of Construction, Hospitality, and Transportation. Every sector roughly doubled adoption 2021→2024; the absolute gap between leading and lagging sectors widened (ICT pulled away by ~15 pp).
Sectoral AI-intensity decomposition (Calvino et al. 2024)
Four-dimensional AI-intensity indicator across 38 A38 sectors (manufacturing ex. coke/petroleum, construction, business services). Top-quartile (dark navy) across all four indicators (human capital, innovation, exposure, use):
- Media (58–60)
- Telecommunications (61)
- IT services (62–63)
- Computer & electronics (26)
- Scientific R&D (72)
Top-quartile on three or more:
- Finance & insurance (64–66)
- Legal & accounting (69–71)
- Other business services (73–75)
Bottom-quartile (white / lightest) across most indicators:
- Food products (10–12)
- Textiles & apparel (13–15)
- Wood & paper (16–18)
- Hotels & food services (55–56)
- Construction (41–43) — sometimes second quartile on use
Mid-tier (mixed):
- Pharmaceuticals (21) — top on human capital; lower on exposure
- Electrical / machinery / transport-equipment manufacturing — second to third quartile
Source: Calvino et al. (2024), reproduced in oecd-sme-ai-adoption-2025, Figure 5, p. 14.
G7 stratification
OECD reports that in G7 economies, ICT sector adoption is typically more than 3× manufacturing sector adoption. This holds for all G7 countries; per-country detail in the source’s Table A.2 / Annex.
Caveats
- “Sector” maps onto NACE / ISIC categories; comparability across statistical agencies is approximate.
- AI definitions broadened over 2021–2024 (generative AI now explicitly captured).
- The Calvino et al. AI-intensity index is a composite — different weighting choices produce different rank orders, particularly in the middle of the distribution.
Why this matters
- Sectoral concentration compounds the sme-ai-adoption-gap. SMEs are over-represented in low-AI-intensity sectors (construction, hospitality, transport, retail) — so part of the size gap is a sector compositional effect. OECD work shows the size gap persists within sector, but the sectoral pattern multiplies it.
- The AI-producing sectors are the AI-adopting sectors. ICT, Media, IT services lead on both innovation and use — the part of the economy that makes AI is the part that captures most surplus today. Consistent with capital-labor-divergence.
- The diffusion question is at the lagging sectors. Per diffusion-vs-innovation, the GPT winner is whoever absorbs the technology across all sectors. The pattern shows OECD economies are still early on the absorption curve for construction, manufacturing, transport, hospitality.
Related
- oecd-sme-ai-adoption-2025 — primary source.
- sectoral-ai-diffusion-pattern — concept page.
- sme-ai-adoption-gap — the inter-firm gap.
- diffusion-vs-innovation — the diffusion-S-curve framing.
- H2_u-curve-of-value — application-side surplus is concentrated in AI-intensive sectors.
- capital-labor-divergence — surplus flow from low-intensity to high-intensity sectors.
- deloitte-ai-dossier-eri — vertical-industrial counterpart for mining / energy.