Use Cases

Do 8% or 41% of Spanish Businesses Use AI? The Data Contradicts Itself

Critical analysis of public figures on AI adoption in Spanish businesses (2025-2026). Official sources (INE, Eurostat, IndesIA) versus sponsored reports inflating the percentages.

Equipo Utilia
10 min
#AI adoption #Spanish businesses #statistics #INE #Eurostat #real data #digital transformation #critical analysis
Do 8% or 41% of Spanish Businesses Use AI? The Data Contradicts Itself

Why nobody agrees

You have been reading posts for two weeks on AI adoption in Spanish businesses. One headline says “41% of SMEs already use AI daily”. The next, published the same day in another outlet, says “only 8% of Spanish SMEs have adopted AI”. A third, quoting the Bank of Spain, mentions 20% for all of Europe.

Who is lying? Nobody. Almost all the figures are technically correct. What changes is what is measured, who asks, and what counts as “using AI”. This post puts the six serious sources on the table, explains the methodologies and gives you a tool so that next time you know which number to trust.


The six figures circulating, with their methodology

These are the public figures circulating in May 2026 about AI adoption in Spanish businesses:

FigureSourceYearSampleWhat it counts as “using AI”Company sizeReliability
2,9%IndesIA, AI in Spanish SMEs Barometer 20252024 / pub. 05/202568.000 SMEs, analysis on real data (not survey)Productive implementation detectable: ML, automation or generative AI in recurrent useSMEsHigh. Only one based on observation, not self-declaration
~17%INE / Eurostat (ICT-EC survey)Q1 2025 / pub. 2025~15.000 companies, stratified samplingUse of any of 8 AI technologies (text, voice, ML, image, RPA…)Small (10-49 employees)Very high. Official harmonised EU statistic
21,1%INE, ICT and e-commerce surveyQ1 2025 / pub. 10/2025~15.000 companiesSame Eurostat criterionCompanies of 10+ employees (includes large)Very high, but not an SME figure
23,3%Hiscox, II SMEs & Self-employed Report2025 / pub. 12/2025Not specifiedSelf-declaration of useSMEs + self-employedMedium. Insurer, declarative survey
41%YouGov for IONOSQ1 2026 / pub. 05/2026514 SMEs up to 250 employeesSelf-declaration of “daily use”SMEs up to 250 employeesLow-medium. AI hosting provider measuring its market, small online sample
56%Sage / Strand Partners, AI Action Summit2025 / pub. 09/2025Not specifiedSelf-declaration of using AI toolsSMEsLow. AI software provider, no public methodology

As a bonus, two more figures often misquoted are worth keeping in mind:

  • 20%: EU-27 average (Eurostat, December 2025), not Spain. Often misquoted as a national figure.
  • 84%: Salesforce State of Sales 2026. Measures sales professionals who use AI, not SMEs.

Why the numbers contradict (and almost all can be true)

Four factors explain the confusion:

1. What is measured

A salesperson using ChatGPT once a month is not the same as having a scoring model in production. IndesIA observes real adoption: it needs evidence, not self-declaration. That is why it gives 2,9%. IONOS asks perception: how many SMEs say “I use AI daily”. That is why it gives 41%. Both can be true; they simply measure different things.

2. Who runs the survey

A useful rule of thumb: when the study sponsor sells AI or related services, figures tend to be inflated. IONOS sells hosting with AI. Sage sells business software with AI. Salesforce sells a CRM with AI. It is reasonable that their surveys detect optimistic adoption in their target market.

When the study is done by a public body or neutral institute (INE, IndesIA founded by Repsol-Telefónica-Banco Santander with educational purpose, Eurostat, Bank of Spain), figures drop radically.

3. Who answers

An online survey to “business decision-makers” with samples of 500-1.500 people overrepresents digital companies and enthusiastic respondents. INE stratifies its 15.000 questionnaires over the real business landscape, including non-digital companies. The sampling matters.

4. Company size

Eurostat published in December 2025 a key disaggregation for the EU:

  • Small companies (10-49 employees): around 17% under the 2025 Eurostat criterion for the EU-27 (approximately 17,9% in Spain per INE Q1 2025).
  • Medium companies (50-249 employees): around 21%.
  • Large companies (250+ employees): 49,2% use AI.

When you see a headline about “Spanish companies” without breakdown, it is usually averaging the three segments, producing a number that looks big but does not represent a typical SME. Spain for 10+ companies sits at 21,1% (Q1 2025), vs the 20% EU-27.


The reasonable reading

If your business has between 5 and 50 employees and you want to know where you stand in the market, this is the honest reading of the May 2026 data:

  • Real adoption with AI integrated in productive processes: 2,9% according to IndesIA’s direct observation. The figure rises to ~17% if any AI technology (including occasional use) is counted, per Eurostat/INE criteria for small enterprises.
  • Sporadic or experimental use (some employee uses ChatGPT, Copilot, etc.): probably between 23% and 41%, but without structure or measurement (Hiscox, IONOS).
  • EU-27 average for companies 10+: 20%, with Spain slightly above (21,1%).
  • Year-on-year growth: Eurostat reports moving from 13,5% (2024) to 20% (2025) in the EU, +6,5 points in a year. Spain rose from 12,3% to 21,1% in 10+ companies (+8,8 points).

That growth is the important news, not the snapshot. Adoption is doubling in two years.


How to read the figures next time

Four questions to ask before citing an AI adoption number:

  1. Who funds the study? If it is a provider selling AI, adjust expectations downward.
  2. Is it observation or self-declaration? Observation gives smaller but more reliable numbers.
  3. What company size does it cover? Without breakdown by employees, the data tells little about an SME.
  4. What does it count as “using AI”? With a loose definition (any tool with an algorithm), the percentage rises. With a strict one (models in production with metrics), it drops sharply.

Applied to your case: if you want to know where you stand vs comparable competitors, look at IndesIA and Eurostat by size. If you want to understand the speed of the market and plan, look at Eurostat and INE year-on-year evolution.


Practical implications for a business

That real adoption hovers between 3-9% in small businesses means two things:

  1. The competitive window remains open. Whoever implements AI with judgement before the curve accelerates (likely 2026-2028) gets a differential advantage. It is not a saturated race yet.
  2. The PR noise does not reflect operational reality. Almost nobody in a typical Spanish business has an AI system in production. The success stories published are the exception, not the average. If your company has not started yet, you are not “late”.

To understand what types of projects are delivering measurable ROI in real Spanish businesses, see the specific analysis with cases in our post on ROI of AI in SMEs: real cases and numbers.


Investment and return figures (the ones with consensus)

There is one area where figures are more stable, because they come from accounting projects rather than surveys:

  • Typical initial investment in a Spanish business: between 3.000 and 50.000 € depending on scope. One-shot projects (chatbot, automation of one process) between 4.600 € and 19.200 €.
  • Typical first-year ROI: between 3x and 9x depending on the application area. Consulting and professional services usually see high returns from hours freed up (15 hours per employee per week on average).
  • Success rate: depends heavily on scoping. Projects with clear delimitation and metrics exceed 70% goal completion. Generic “implement AI” projects fail in more than 50% of cases.

For a more detailed breakdown of real costs, see our post on how much does AI cost for an SME.


Frequently asked questions


Sources

Was this article helpful?

Discover how we can help you implement these solutions in your company