Economic and industry statisticsDigital infrastructure and AI
INE survey puts business AI use at one third of Spanish firms
Spain’s statistics office reports that 33.6% of firms with at least 10 employees used AI in the first quarter of 2026, alongside broad uptake of paid cloud services.
- Event
- First published on EPC Feed
- Sources
- 3 listed below
- Reading time
- 3 min
AI adoption, alongside cloud and online sales
Spain’s National Statistics Institute (INE) reported on 22 October that 33.6% of businesses with at least 10 employees used artificial intelligence during the first quarter of 2026. The same release put paid cloud computing use at 39.4% among these firms. The figures come from the definitive edition of the annual Survey on the Use of Information and Communication Technologies and Electronic Commerce, covering 2025 and the first quarter of 2026. The event date here is the publication date of the INE release; the measured technology use refers to the first quarter, while e-commerce activity refers to 2025.
The headline adoption rate masks variation by size and sector. Among firms with 10 or more employees, 98.3% had an internet connection, and 82.9% of connected firms had a website. The share using social media was 65.4% of connected firms. Services companies reported the strongest use of AI among the broad sectors shown, at 38.2%, compared with 29.1% in industry and 22.6% in construction. Paid cloud use followed a similar pattern: 41.9% in services, 37.5% in industry and 32.9% in construction. These percentages describe adoption, not intensity, spending, productivity gains or the business purpose of each system.
The survey also distinguishes small firms. Among companies with fewer than 10 employees that were connected to the internet, 21.4% used AI in the first quarter of 2026. Across all firms in that size group, 81.2% had an internet connection. Among those connected, 25.1% allowed teleworking and 36.2% had a website. The comparison is useful because a single national percentage would conceal a substantial size gap, but the two populations are reported in separate survey sections and should not be treated as identical denominators.
A mixed reference period matters
Electronic commerce is measured against the preceding calendar year: 29.0% of businesses made sales through e-commerce in 2025, generating €419.706 billion in turnover. That is a different measure from firms’ first-quarter 2026 use of AI and cloud. The release also reports that 15.4% of firms with at least 10 employees employed ICT specialists in the first quarter, while 6.6% employed women ICT specialists. These data describe workforce and infrastructure context, but they do not establish that firms using AI also employed specialist staff or made online sales.
INE describes the operation as a continuous annual survey covering specified sectors of the economy, across Spain. Its methodology says technology-use variables refer to the first quarter of the survey year, while general company information, e-commerce and ICT training refer to the prior year. The collection is multichannel, including online questionnaires and postal mail, with stratification by business size, activity and autonomous community. Those design details explain why the release reports multiple populations and reference periods. They also counsel against reading the headline AI share as a real-time measure or a causal account of digital investment.
INE also flags a break in comparability: from reference year 2025 the survey uses the CNAE-2025 activity classification and expands its population to additional activity groups. Its definitive 2025–2026 results are therefore not strictly comparable with prior-year releases. The published adoption levels should be interpreted within this edition’s population and definitions rather than treated as a like-for-like growth rate. The release does not report a causal link between AI adoption and company productivity. The survey measures whether technologies were used; it does not assign an investment return to them.
Sources
Spotted an error or have a primary source we should read? Send a correction.