Normandie Digital

L'actualité du web, du référencement et de l'IA, expliquée pour les TPE et PME.

🇫🇷 FR

French firms and digital technology: Insee 2025 data

In brief: In 2025, 18% of French companies with 10 or more employees used artificial intelligence in an organised way, three times more than in 2023 (Insee, ICT survey 2025). The France Num 2026 Barometer reports 40% of TPE-PME using a broader definition. The gap is methodological: organised use for Insee, self-declared use for France Num. The main barrier is not cost: 71% of non-users see no use for it.

AI use triples in two years but remains a minority

In 2025, 18% of French companies with 10 or more employees declared using at least one AI technology in their services, according to Insee (Insee Première no. 2120, July 2026). That is 8 percentage points more than in 2024 and 12 more than in 2023. The rate has therefore tripled in two years, after a long plateau.

This 18% remains a minority. Company size matters: use reaches 15% for companies with 10 to 49 employees, 31% for 50 to 249, and 58% for 250 or more. Companies with fewer than 50 employees, which make up 85% of the survey field, remain far behind larger structures.

By size and year: 10 to 49 employees: 5% in 2023, 9% in 2024, 15% in 2025. 50 to 249: 10%, 15%, 31%. 250 or more: 21%, 33%, 58%. Overall: 6%, 10%, 18%. Source: Insee ICT surveys 2023, 2024 and 2025; Eurostat.

France remains slightly behind the European Union (20% of companies using AI), but the gap has almost closed. Insee notes that French companies of the same size as their European counterparts caught up in 2025, except those with fewer than 50 employees, which remain 2 points behind. For context: Germany is at 26% and Nordic countries between 33% and 42%.

Sector matters as much as size. Information and communication companies reach 59% use, according to Insee’s 2025 ICT survey. At the other end, construction (10%), accommodation and food service (12%) and retail (13%) remain at the bottom, even though their rates doubled or tripled in one year.

ERP, CRM and cloud: the management base

From detailed tables of the 2025 ICT survey (representative sample of 194,000 companies), equipment rates by size can be calculated. For companies with fewer than 20 employees: ERP 45%, CRM 25%, paid cloud services 33%. For companies with 250 or more employees: ERP 93%, CRM 74%, paid cloud services 85%. Calculations from Insee company counts (TICE 2025 tables, field: 194,243 companies with 10 or more people).

Across the whole field, 54% of companies use an ERP, 31% a CRM and 40% paid cloud services. The gap between companies with fewer than 20 employees and those with 250 or more therefore reaches 48 points for ERP. A small-business manager often works with a spreadsheet and email, while large companies have shared data in a single system for years.

This base matters for what follows, because Insee observes a clear link between cloud and AI: 73% of companies that use AI also use paid cloud services, against 33% of those that do not. Adoption of advanced technologies relies on infrastructure already in place, not the other way round.

One company in six sells online

The 2025 ICT survey also asks about web sales (own sites and marketplaces). In 2024, the latest measured year, 17% of companies in the field made web sales of goods or services, according to Insee’s 2025 ICT tables. The share of very small structures (fewer than 20 people) is lower still, at 13%.

This figure contrasts with the impression given by talk of digitalisation. Online commerce remains a marginal channel for most companies with 10 or more employees, even two years after the post-Covid boom. Sector detail confirms this: construction has almost zero online sales, while retail and accommodation and food service pull the statistic up.

For a very small business still hesitating, the data does not say not to sell online. It mainly says the channel has not become a general obvious choice, and that companies entering it do so in specific sectors.

18% for Insee, 40% for France Num: why such a gap

An attentive reader will find an apparent contradiction. The France Num 2026 Barometer reports that 40% of TPE-PME use at least one artificial intelligence solution (source: France Num 2026 dossier, DGE press release no. 1062). Insee reports 18%. Two official sources, two figures, on the same subject and the same year.

The gap is explained by methods, and it must be understood before citing either figure. Insee, ICT survey 2025: field of companies with 10 or more employees (194,000 companies); measures organised use of AI within services, list of 8 precise technologies; AI result 18%; strength: European comparability, stable series since 2022. France Num 2026 Barometer: field of TPE and PME, micro-enterprises included (Crédoc sample of 9,655 companies); measures self-declared use of AI solutions, broader definition; AI result 40%, against 13% in 2024; strength: covers micro-enterprises, absent from the Insee survey.

Insee also states in its methods that answers do not include occasional and individual uses by employees, such as using a chatbot. Yet this is precisely the type of use a very small business manager declares spontaneously in a telephone survey. Both figures are therefore true within their own framework: 40% measures declared familiarity, 18% measures organised adoption.

The same reading applies to the progression: France Num goes from 13% to 40% in two years, Insee from 6% to 18%. The trajectories are parallel and give the same information, a tripling, with levels shifted by definitions.

Barriers and uses

Among companies that do not use AI, 71% say they see no use for it in their activity, and 54% cite a lack of expertise, again according to the 2025 ICT survey. High costs come only sixth, with 32% of responses. The barrier is first cognitive and organisational, not budgetary.

This result should be read with nuance. “No use” is the default answer from a manager who has not yet identified a use case in their trade, rather than a definitive judgment. Insee also observes that 53% of companies already using AI say they are held back by a lack of expertise, and 43% by concerns about data protection. Barriers do not disappear after adoption; they change in nature.

The declared purposes of users give an idea of the use cases actually tested. IT security comes first (39% of affected staff), ahead of organising administrative processes (38%, up 21 points in one year) and research and innovation (35%). Marketing, despite being the most publicised subject, comes only fourth (31%). And 80% of AI-using companies buy ready-made software rather than developing in-house.

What this means for a very small business

First reading: small companies are not behind on everything. France Num shows strong familiarity with AI, and Insee confirms rapid progress everywhere. The delay lies in structuring: ERP at 45%, CRM at 25%, paid cloud at 33% among companies with fewer than 20 employees. This base, more than AI itself, drives the gap.

Second reading: the order of priorities often reverses the order of public discourse. Companies using AI mobilise these technologies first for security and administration, not marketing. The majority path is therefore commercial off-the-shelf tools, not bespoke technology projects.

Third reading, the most honest: these statistics describe national averages, not a roadmap. A transport company with 15 employees and a communication studio with 30 do not have the same needs or the same room for progress. The Insee survey is better read as a benchmark than as a target to reach.

Sources

Raphaël Marieux Mis à jour le 13/11/2026

J'explique le web et l'IA en français, sans jargon ni pub.

Tous ses articles