The Swiss AI paradox: world champion in talent, late adopter among SMEs

TLDR: Switzerland holds the highest density of artificial-intelligence specialists in the world, while organisational adoption stays modest among small and mid-sized businesses: the brake is structural, rooted in how companies are organised more than in the available talent.

One small Swiss firm in twelve uses AI, against one large firm in three

Artificial intelligence (AI) advances through the Swiss economy at two very different speeds. According to the KOF Swiss Economic Institute at ETH Zurich, just over 8 percent of small firms use AI, against roughly 34 percent of large ones. The gap widens further on data: 20 percent of small structures work with big data (the analysis of large data volumes), against 60 percent of large firms. Where the two meet, the picture turns stark: fewer than 5 percent of all firms apply AI to the analysis of their own large data volumes.

This split by size tracks a documented European pattern, where Eurostat records 12.7 percent adoption among small enterprises against 41.2 percent among large ones, and where the OECD observes a ratio of 12 percent to 39 percent. Organisation size, everywhere, dictates the depth of use.

The paradox: the country counts the most AI specialists in the world

The contrast becomes striking when set against human capital. The Stanford AI Index 2026 ranks Switzerland first worldwide for the density of AI specialists, with 110.5 experts per 100,000 inhabitants, ahead of Singapore (109.5), Sweden (80.6) and the United States (64.8). On individual use, the Federal Statistical Office (FSO) places Switzerland among the top European ranks for the share of residents using generative AI, nearly half of residents, behind Norway and Denmark.

The talent sits ready, and so does mainstream familiarity. The weak link lives between the two: the conversion of that abundance into value at company scale. The shortfall among SMEs (small and mid-sized enterprises) reflects the capacity of organisations to capture, structure and deploy skills that the country already holds, against their own processes.

The real gap turns on depth of use, beyond the number of tools

The adoption curve climbs fast all the same. The AXA Labour Market Study 2025, relayed by the Confederation’s SME Portal (SECO), shows AI use in Swiss SMEs rising from 22 percent to 34 percent in a single year, the share of leaders who stayed clear of these tools falling from 45 percent to 29 percent, and 57 percent now recording efficiency gains. The most common use cases stay shallow, however: translation (52 percent), correspondence (47 percent), automation (34 percent) and data analysis (32 percent).

That distribution tells the essential story. An SME that translates an email consumes a commodity available to every competitor. An SME that queries its own data builds a defensible advantage. The AWS study “Unlocking Switzerland’s AI Potential” sizes the step to clear: 62 percent of Swiss firms confine AI to basic tasks, and only 19 percent of the SMEs that have adopted it build products or services driven by AI. Depth separates cosmetic use from structural advantage; the count of software subscriptions moves that line very little.

Data governance stays the silent liability of adoption

Beyond depth, a compliance risk settles in quietly. Still according to the AXA study relayed by SECO, only 34 percent of SMEs have set rules governing the data fed into AI tools, and that share drops to 23 percent for structures under ten employees. A majority of firms therefore let customer, contractual or strategic information flow through third-party systems with a defined framework absent.

The barriers identified converge on the same organisational root. The HWZ-Swisscom study puts the shortage of qualified skills at the top, alongside regulatory uncertainty, haziness over return on investment and weak data quality. The KOF adds high cost as the leading obstacle to innovation. These barriers belong to decision and internal structuring, two levers that depend on leadership far more than on the labour market.

Large groups pull ahead by industrialising use

Where large organisations widen the gap, they do it by moving from the one-off tool to a framed deployment. The Adecco group offers a documented illustration: according to a Microsoft case study, its recruiters gained up to 63 percent in productivity through a generative-AI rollout embedded in their workflows. The technology stays open to every firm; industrialisation makes the difference, through a use case that is framed, measured and generalised.

Swiss SMEs prove that depth stays within reach beyond large groups. Andermatt Biocontrol Suisse has used generative AI for its translations, summaries and reports since 2024 and is building its own conversational agent, while placing the AI-plus-big-data combination on a more distant horizon, given limited internal know-how, investment and data quality. By contrast, the software publisher Cudos AG, around sixty staff, has deployed AI at company scale with a secure internal assistant that queries its data and a project-tracking copilot. Cudos embodies exactly the depth that the majority lacks.

What closes the gap: a leadership priority and a few measured cases

The shortfall being organisational, its resolution is too. Three levers stand out from the data. First, a priority carried by leadership, since the barriers belong to budget arbitration and internal structuring. Second, a limited number of use cases deployed with rigour on the company’s own data, in place of a proliferation of generic tools. Third, a data-governance framework set before use, ideally ahead of any incident.

The canton of Vaud offers a concrete model of that shift. The “Industrie Augmentée / La Cohorte” programme, run by the Canton of Vaud, supports industrial-SME leaders over three months to turn a concrete problem into a deployed AI solution, with a first group of seven entrepreneurs. It sits within the INNDUSTRY ecosystem from Innovaud, which addresses the roughly 5,000 industrial SMEs in the canton. This cohort approach, centred on a measurable case, renders the general thesis into a local method: Swiss talent stays abundant, and company-scale adoption remains the layer still to build. Operators such as Pupsic read the same conclusion, where small businesses gain most from a single measured case run well on their own data.

References

1. KOF, ETH Zurich. Digital technology: are small firms being left behind? 20 February 2025. https://kof.ethz.ch/en/publications/kof-insights/articles/2025/02/digital-technology-are-small-firms-being-left-behind.html
2. SECO, Confederation SME Portal. AI gains ground among Swiss SMEs (AXA Labour Market Study 2025). 5 November 2025. https://www.kmu.admin.ch/kmu/en/home/new/news/2025/ai-gains-ground-swiss-smes.html
3. AXA Switzerland. SME Labour Market Study 2025: artificial intelligence. 8 October 2025. https://www.axa.ch/fr/aproposdaxa/medias/communiques-presse/derniers-communiques-presse/2025/20251008-etude-sur-le-marche-du-travail-pme-2025-ia.html
4. Stanford HAI. AI Index Report 2026. April 2026. https://hai.stanford.edu/ai-index/2026-ai-index-report
5. Federal Statistical Office (FSO). ICT Omnibus 2025: internet and artificial-intelligence use. December 2025. https://www.bfs.admin.ch/bfs/fr/home/actualites/quoi-de-neuf.assetdetail.36304418.html
6. AWS. Unlocking Switzerland’s AI Potential 2025. https://www.unlockingeuropesaipotential.com/switzerland
7. HWZ Hochschule für Wirtschaft Zürich, Swisscom. Einblick in die Nutzung von künstlicher Intelligenz in Schweizer Unternehmen. 5 February 2025. https://fh-hwz.ch/de/news/medienmitteilungen/einblick-in-die-nutzung-von-kuenstlicher-intelligenz-in-schweizer-unternehmen
8. Microsoft. Adecco Group AG: Microsoft 365 Copilot. 10 July 2025. https://www.microsoft.com/en/customers/story/24691-adecco-group-ag-microsoft-365-copilot
9. Canton of Vaud. Industrie Augmentée: 3 months to activate AI in your SME. 13 October 2025. https://www.vaud.ch/actualites/industrie-augmentee-3-mois-pour-activer-lia-dans-votre-pme/
10. Innovaud. INNDUSTRY programme. https://www.innovaud.ch/fr/programmes/nos-programmes/inndustry
11. Eurostat. Use of artificial intelligence in enterprises. 23 January 2025. https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20250123-3
12. OECD. AI adoption by small and medium-sized enterprises. December 2025. https://www.oecd.org/en/publications/ai-adoption-by-small-and-medium-sized-enterprises_426399c1-en.html