Europe’s AI uptake grows, but productivity gains remain elusive

AI adoption is expanding rapidly across Europe, but it is still too early to detect a material improvement in aggregate productivity, Philip R. Lane said in remarks published on 13 October 2026. Firms are expected to allocate about 10% of their total investment to AI this year, while AI-related lending represents roughly a quarter of recent growth in credit to businesses.

In “The AI opportunity for Europe”, published by the European Central Bank, Lane described a technology already influencing investment, employment prospects and financial conditions—even before its longer-term productivity benefits become measurable. The speech draws on firm and household surveys, alongside research on international adoption and financial risks. It is available at https://www.ecb.europa.eu/press/key/date/2026/html/ecb.sp261013~9db9bf4fd6.en.html.

Widespread use is not yet intensive use

The Survey on the Access to Finance of Enterprises for the fourth quarter of 2025 found that 64% of firms used AI experimentally or moderately. Only 7% described their use as significant—a distinction that matters when assessing how far adoption has translated into changes in production.

Household survey findings cited by Lane show that 52% of workers now report using AI at work, up from 26% in 2024. Workers report substantial time savings overall, but those gains are concentrated in particular industries and specialised tasks, including coding. These are reported experiences, not a measured economy-wide productivity increase.

Among firms planning AI-related investment over the following 12 months, 49% intended to spend on technologies and tools, 46% on employee training, 40% on data and infrastructure, and 12% on hiring AI specialists. The plans were expected to be financed mainly from internal funds; the categories represent different spending intentions rather than shares of a single investment budget.

Employment effects depend on how work changes

Lane said aggregate employment trends had not yet changed substantially, although subdued hiring plans were particularly affecting prospects for new graduates. Survey responses also differed by industry: workers benefiting from AI infrastructure construction were more optimistic than those in knowledge-intensive sectors exposed to substitution.

He identified weaker worker bargaining power, slower wage growth and a reduced labour share of income as possible consequences—not established outcomes. Whether AI ultimately increases or reduces employment will depend partly on which tasks it complements, which it replaces, and the relative cost of labour and AI processing.

The broader transition requires more than stronger models. Lane pointed to energy supply, financing, training, infrastructure and redesigned business processes as interdependent requirements. A constraint in any one area can delay the adoption of advances made elsewhere.

Europe trails the scale of the global boom

Over the preceding two years, US digital investment grew twice as fast as euro area digital investment, according to the comparison Lane cited. US workers also spent two to three times as much of their working week using AI as workers in the largest euro area economies. Digital investment is a broader measure than AI spending alone.

Lane cited a Stanford AI Index count of 59 notable models produced by the United States and 35 by China, against one each for France and the United Kingdom. He also put the US share of global AI computing capacity at 75%, compared with Europe’s 5%. The remarks do not specify the reference period for those model and computing-capacity comparisons.

International trade can distribute the benefits of AI goods and services, he argued, but unequal national capabilities can also create economic and geopolitical difficulties. For Europe, higher productivity could be especially valuable in cushioning the effects of a projected decline in its working-age population.

Global investment can tighten European financial conditions

Lane suggested that heavy global AI investment and expectations of future productivity growth were plausibly contributing to higher long-term interest rates. This was an assessment of possible causes, rather than a quantified estimate of AI’s contribution.

Because Europe’s own AI expansion is smaller, the transmission of higher global yields into euro area bond markets can tighten its financial conditions without a comparable domestic investment boom. Lane identified that imbalance as relevant to monetary policy decisions.

He also warned that elevated AI company valuations and growing AI-related debt issuance could create financial instability if technological progress disappoints—even temporarily—or investor confidence falls. Frontier companies’ reliance on equity financing and, increasingly, private credit makes their funding sensitive to global appetite for risk.

The central bank’s financial stability and banking supervision teams are monitoring those exposures alongside cyber risks. Lane also described an internal AI adoption programme intended to improve analytical work and organisational efficiency, while stressing that Europe’s transition needs both price stability and financial stability.