OECD Finds 52% of Large Firms Use AI, More Than Twice the 20% Rate Across All Firms
Large firms in the OECD are more than twice as likely as the average firm to use AI, 52% against 20%, a gap projected to persist without targeted support.
For OECD governments, the new Outlook shows AI concentrated where resources already sit: in large companies, among younger and better educated adults, and in the infrastructure layer that venture investors fund. Smaller firms remain constrained by limited access to finance, digital infrastructure and specialised skills, the report found, and the 20% average covers enterprises with 10 or more employees. Venture investment in AI infrastructure firms more than doubled in 2025, while four in five of those enterprises still did not use AI; in The Edge’s reading, the money is running ahead of the users.
The OECD Digital Economy and Society Outlook 2026, published on 5 November, covers firm and household adoption of digital technologies, connectivity, investment in AI compute and the digital policies of 42 countries.
Large firms stay ahead on AI as cloud use converges
AI was used by 20% of enterprises with 10 or more employees across the OECD in 2025. Among large enterprises, those with 250 or more employees, the share was 52%. The gap is 32 percentage points, and large firms are 2.6 times as likely to use the technology.
Cloud computing shows the opposite path. More than half of all firms and over 80% of large firms used it in 2025, and the report’s projections have smaller firms reaching near comparable levels by the end of the decade.
AI adoption is not projected to converge. Large firms have consistently led AI uptake since 2020, and by 2030 their adoption could be nearly 80% higher than the average across all firms, according to the projections. Implementation costs, uncertainty over legal and data protection rules and the difficulty of fitting AI into existing systems add to the finance and skills constraints.
OECD Secretary-General Mathias Cormann said “we need to make sure adoption spreads to firms of every size”, pointing to wider access to cloud services, targeted support for AI and big data analytics, and stronger interoperability and open standards.
Firms mainly apply AI to marketing, business administration, finance and ICT security. Combined use of AI, cloud computing and big data analytics stays concentrated in large firms and knowledge-intensive sectors; the highest shares of firms using all three, above 20%, are in Belgium, Denmark, Finland, Korea, the Netherlands and Sweden.
Generative AI reaches 37% of adults, with age the widest divide
Generative AI use among individuals reached 37% in 2025, with higher uptake among younger, highly educated and higher income groups. Disparities are greatest by age.
Among 16 to 24 year olds, 68% used generative AI. The share among adults aged 55 to 74 was 14%, making the young 4.9 times as likely to use it.
Students aged 16 and over reached 76%. After them, use was highest among the employed (41%) and the unemployed (37%).
Older adults are closing the internet gap but not the AI gap. Internet use among 55 to 74 year olds rose to 87% in 2025 from 58% a decade earlier. That gain of 29 percentage points has narrowed basic access.
Online public services still show a shortfall of 13 percentage points. In 2025, 68% of older adults used them, against 81% across all ages.
Among older adults, 43% saw no need for AI in 2025, almost double the 25% among 16 to 24 year olds. Smaller shares cited a lack of skills (12%) or of awareness (9%).
AI and generative AI use across the OECD, 2025 (in %)
| Group | Share |
|---|---|
| Firms with 10 or more employees | 20 |
| Large firms, 250 or more employees | 52 |
| Adults using generative AI | 37 |
| Aged 16 to 24 | 68 |
| Aged 55 to 74 | 14 |
| Students aged 16 and over | 76 |
Firm rows show the share using AI; adult rows show the share of people aged 16 to 74 using generative AI tools to create content in the last three months; 2025 or the most recent year available.
Venture money into AI infrastructure more than doubles
Venture capital investment in AI firms working on IT infrastructure and hosting rose to $123.9 billion in 2025 from $54.5 billion in 2024. The $69.4 billion increase is a rise of 127.3%.
The report puts the same category at an estimated $30 billion in 2023, and says infrastructure and hosting accounted for almost half of all AI venture capital investment in 2025. Set against a 2023 estimate of $54 billion of investment in AI-related computer hardware in the EU27, Canada, the United Kingdom and the United States, that figure suggests total spending on AI-specific compute is significantly higher than venture or private equity investment, the report says.
Physical limits are emerging alongside the money. Citing the US Department of Energy and Data Center Frontier, the report noted that hyperscale data centre buildouts can take up to three years while securing multi-gigawatt power capacity can take more than four, a gap it says could create bottlenecks to AI progress.
Venture capital in AI infrastructure and hosting firms (in $ billions)
| Year | Investment |
|---|---|
| 2023 | 30 (estimate) |
| 2024 | 54.5 |
| 2025 | 123.9 |
The 2023 value is the report’s approximation; 2024 and 2025 are the reported totals.
The wider ICT sector across the OECD is estimated to have grown 5% in 2025. Value creation and sourcing remain highly concentrated, with a small number of countries acting as the main production and supply hubs.
Cybersecurity climbs to second among digital policy priorities
Digital government has ranked as the top digital policy priority in every OECD data collection since 2016. In 2025 cybersecurity rose to second place, its highest position in the decade these data cover, and technology development tied for third with connectivity and skills.
The policy findings draw on 1,497 initiatives identified across 42 countries in 2025. An estimated 40% focus on effective use of digital technologies, the OECD said; digital innovation accounts for 30% and societal issues for 28%.
AI was the most frequently cited technology in policy initiatives in 2025, while quantum technologies, semiconductors and satellites have become more prominent since 2023. Skills is the most prominent policy domain.
Responsibility is moving to specialist ministries. Among countries with a national digital strategy, 60% gave primary responsibility for developing it to a dedicated digital ministry in 2025, up from 24% in 2016. Strategies developed above ministerial level, such as in a prime minister’s office, presidency or chancellery, fell to 26% from 42% in 2023.
The networks underneath are faster. Fibre made up 51% of fixed broadband subscriptions in the fourth quarter of 2025, up from 31% five years earlier. Median fixed broadband speeds reached 207.5 Mbps in the second quarter of 2026, 3.8 times the level of the second quarter of 2020.
Rural areas still trail. City users had median fixed download speeds 42% higher than rural users in the second quarter of 2026. Investment in submarine cables is expected to exceed $4 billion a year over 2025 to 2027, double the annual average from 2016 to 2022.
Digital governance and access, earlier and latest readings (in %)
| Indicator | Earlier | Latest |
|---|---|---|
| Strategy led by a digital ministry | 24 (2016) | 60 (2025) |
| Strategy led above ministerial level | 42 (2023) | 26 (2025) |
| Fibre share of fixed broadband | 31 (Q4 2020) | 51 (Q4 2025) |
| Internet use, aged 55 to 74 | 58 (2015) | 87 (2025) |
Strategy rows cover OECD Member countries and partner economies with a national digital strategy; fibre and internet use are OECD averages.
Figures are from the OECD press release and the OECD Digital Economy and Society Outlook 2026 report, both dated 5 November 2026; ratios, differences, point changes and percentage changes not attributed to a source are The Edge’s calculations.
Sources: OECD, The Edge.

