El uso de la inteligencia artificial por parte de las empresas españolas: avances recientes y factores asociados a su uso (Documento Ocasional 2624)
A · Strong evidence Banco de España occasional paper by Jacob Hess and Diana Posada Restrepo combining a repeated official business survey with balance-sheet records; comparable 2024-2026 data and a clear method, though adoption and effects are self-reported and the links are not causal.
Summary based on the full text·Original: about 30 minutes to read
In short
A Banco de España occasional paper (No. 2624, October 2026), written in Spanish, on how Spanish firms use AI. Using the AI module of the bank's business activity survey for Q1 2026, compared with 2024, it finds adoption has roughly doubled among smaller firms and reached about 70% of large ones, with use still mostly experimental in SMEs. Effects on employment so far are small; reported productivity and investment gains rise with intensity of use. Adopters are more productive, pay more and hold more intangible assets.
Key numbers
- 30-35%small and mid-sized firms using predictive or generative AI in Q1 2026, roughly double 2024 (+16 pp)
- about 70%large firms using AI in 2026, up about 17 pp
- about 89%surveyed firms reporting no significant effect of AI on headcount over three years (about 6% cut, 6% added)
- 44%large firms expecting fewer jobs in repetitive or routine tasks, against 83% of SMEs expecting no change
- 61%non-adopters (N = 2,684) citing resource constraints, high cost or lack of skilled staff, as their main obstacle
As reported by the source.
English·Web page·opens bde.es
What it reports
- SMEs: 30-35% use AI, up 16 pp since 2024; generative AI growth is split between sporadic use (+8.4 pp) and moderate or high use (+8.5 pp). Large firms: about 70%, with +14 pp in moderate or high use.
- Information and communications leads at almost 60% (about 40% in 2024); professional and scientific activities about 50%; agriculture and construction rose about 20 pp to nearly 30%. The sector gap widened.
- Main reasons for use: process quality or reliability (32.5% predictive, 30.7% generative), automating tasks previously done by staff (21.5%, 27.0%).
- Headcount: about 89% report no significant effect; about 6% each cut or added staff. Among heavy generative-AI users, about 13% added staff and about 10% cut.
- More than 60% of heavy generative-AI users report higher productivity, against about 14% of sporadic users.
- Next three years: 61% of large firms expect productivity gains versus 27% of SMEs; 44% of large firms expect fewer routine-task jobs, about 20% more analytical jobs. 79% of large firms plan more training and 62% more specialised hiring (SMEs 35% and 17%).
- Lack of skilled staff remains the main obstacle, cited by about half of firms. Among non-adopters: resource constraints 61%, perceived irrelevance 18%, technical readiness 11%, regulatory or ethical concerns 8%.
How they got it
Responses to the AI module of the Banco de España business activity survey (EBAE), Q1 2026, compared with the 2024 module, then linked by tax ID to the bank's integrated balance-sheet database to compare adopters and non-adopters within sectors on total factor productivity, average wage, intangible intensity and debt ratio.
Limits
- Adoption, effects and expectations are self-reported.
- Balance-sheet data lag the 2026 survey responses.
- Associations only; the authors note productivity may drive adoption as much as the reverse.
- Spain only; size thresholds and sector mix limit comparison with other countries.
Why it matters for companies
For European firms benchmarking themselves, this gives an official reference point: about a third of SMEs and two thirds of large firms now use AI, but most smaller firms are still experimenting. The gains firms report track intensity of use, not adoption as such, and headcount effects remain small. The barriers have moved from cost and legal uncertainty toward skills, which is where large firms are planning to spend on training and hiring.
Key passages in translation
Quoted in English
entre el 30 % y el 35 % de las pequeñas y medianas empresas (pymes) declararon utilizar tecnologías de IA predictiva o generativa, aproximadamente el doble
Between 30% and 35% of small and medium-sized enterprises reported using predictive or generative AI technologies, roughly double.
Quoted in English
la mayoría de las empresas encuestadas declaran que su introducción no ha tenido efectos significativos en los últimos tres años (cerca del 89 %)
Most surveyed firms say its introduction has had no significant effects over the last three years (about 89%).
Quoted in English
la falta de personal cualificado continúa siendo el principal obstáculo
The lack of skilled staff remains the main obstacle.
Summary based on the full text. Read the original for the full argument and data.Read the original on bde.es