Artificial intelligence in business: the support available in Luxembourg

Artificial intelligence in business: where Luxembourg stands

Artificial intelligence in business is no longer a forward-looking topic in Luxembourg. In 2025, 33.6% of companies with 10 or more employees used at least one artificial intelligence technology, compared with an average of around 20% across the European Union, according to STATEC and Eurostat. The rate nearly tripled in four years: it was 13% in 2021, then 23.7% in 2024. Luxembourg companies are adopting AI faster than the European average, which changes the picture for managers who were still waiting to see how things would develop.

Part of the reason for this rapid growth is the size of the Luxembourg market, where lean organisations make decisions quickly, and part is the dedicated public schemes described below. It also creates direct competitive pressure: an SME that has not yet tested anything in artificial intelligence is now behind a third of its peers, and the same source shows the gap widening every year.

For a manager, this figure has an immediate practical consequence. AI is no longer a bet on the future but something a significant share of direct competitors already use, often quietly, through automation or decision-support tools built into software they already run. The growth measured by STATEC and Eurostat over four consecutive years, from 13% in 2021 to 33.6% in 2025, is therefore not a one-off headline but an underlying trend that lets every manager see where their own company stands against the local market.

Why managers adopt AI: productivity, processes, costs

According to a Luxinnovation survey of Luxembourg professionals, 88% of respondents cite productivity gains as the main driver of AI adoption in their company. Process optimisation comes second (65%), followed by cost reduction (64%). These three motivations match the most common uses seen in Luxembourg: automating repetitive tasks, writing assistance, customer data analysis and decision support.

These figures show that AI is used first to free up time in marketing or administrative teams that are already small, and only later as a matter of product innovation. For a Luxembourg SME, the immediate challenge is therefore less about choosing the most advanced technology and more about identifying the most time-consuming task to automate first. Some Luxembourg companies are already going further and testing generative AI directly in their advertising campaigns, as described in our guide to advertising with ChatGPT Ads.

The productivity, processes and costs trio also gives a simple way to prioritise a first project: a manager choosing between several automation ideas can ask which one most directly addresses one of these three levers before applying for funding.

Funding your project: the Fit 4 AI and SME Packages schemes

Luxembourg has set up two complementary schemes to support SMEs financially in adopting artificial intelligence, both run by Luxinnovation.

The Fit 4 AI programme funds an AI diagnostic and roadmap carried out by an approved consultant. The diagnostic is funded at 50% of eligible costs for an SME and 30% for a large company, for eligible costs between 10,000 and 200,000 euros excl. VAT, according to Guichet.lu. This first step maps the company's real needs before any technical investment, which avoids buying a tool that is oversized or a poor fit.

Once the roadmap is in place, the SME Packages - AI scheme, launched in early 2025, takes over for operational implementation: it funds 70% of projects with eligible costs between 3,000 and 25,000 euros excl. tax, according to the Chambre de Commerce du Luxembourg and Luxinnovation. In practice, an SME can have both the diagnostic and the rollout of a first AI tool funded, without bearing most of the cost of either step on its own.

This two-stage design follows a specific logic: preventing a company from investing in a tool before it has clarified its need, then helping it financially once that need has been documented by an approved third party. A manager in a hurry may be tempted to skip the first step and go straight to buying a solution, but then loses access to the 70% funding reserved for implementing a Fit 4 AI roadmap.

The amounts involved show the scale the authorities have in mind: a 10,000-euro Fit 4 AI diagnostic costs an SME 5,000 euros once co-funding is applied, and a 25,000-euro implementation project through the SME Packages leaves 7,500 euros for the company to pay. This level of remaining cost suits projects designed for small organisations, not large transformation programmes.

Submitting your application on MyGuichet.lu: the practical steps

Co-funding applications for both Fit 4 AI and the SME Packages - AI are made through a professional account on MyGuichet.lu, according to Guichet.lu. In practice, you first select an approved consultant for the diagnostic stage, then submit the administrative application online.

The order of the steps matters: the Fit 4 AI diagnostic must be completed before you can claim implementation funding through the SME Packages, since the second scheme funds execution of the roadmap produced by the first. A manager who skips this step risks paying alone for a tool poorly matched to the company's real needs.

For the specialist technical profiles this kind of project sometimes requires in-house, there is also a local talent pool: our article on the EUCYS 2026 contest lists young Luxembourg deep-tech talent that companies still struggle to recruit, even as demand for AI skills grows alongside adoption of these technologies.

Who oversees AI in Luxembourg: the CNPD, AI Act authority

Funding does not remove the need for compliance. A bill approved on 29 November 2024 designated the CNPD as the national competent authority for the AI Act, the European regulation on artificial intelligence, according to a CNPD press release. It acts as the single point of contact, coordinates the other competent authorities and carries out market surveillance by default.

There are two exceptions: ILNAS is designated market surveillance authority for AI systems covered by points 1 to 10 of Annex I and by Article 6 of the European regulation, and sector authorities keep their usual remit, the CSSF for finance and the Commissariat aux assurances (insurance regulator) for insurance, again according to the CNPD.

A Luxembourg manager therefore has an identified point of contact before even deploying a first AI tool, which avoids navigating blindly between several European regulators. For a company that does not operate in finance or insurance, the CNPD is the first port of call for any AI-related compliance question.

Testing before deploying: the Sandkëscht sandbox

On 24 May 2024, the CNPD launched a regulatory sandbox called Sandkëscht, open to any entity established in Luxembourg regardless of size or sector. Participation lasts 9 to 18 months and allows the GDPR compliance of an artificial intelligence project to be tested before it is placed on the market, according to the CNPD.

Sandkëscht is the most concrete tool a Luxembourg SME has to check, before signing a contract with an AI vendor, that its project complies with the GDPR, rather than waiting for an after-the-fact inspection. The CNPD also recommends applying the “privacy by design” principle from the design stage of any AI-based processing of personal data, a principle that is checked precisely during the time spent in the sandbox.

A specific point of attention for HR teams: the CNPD points out that AI tools used for recruitment, appraisal or dismissal must meet the transparency and justification requirements already set by Luxembourg labour law. A CV-screening algorithm or performance-scoring tool therefore falls under the same oversight as an equivalent human decision, with the same obligation to justify the decision taken.

This compliance work overlaps heavily with the data security good practices already recommended to Luxembourg SMEs, set out in our 4-week cybersecurity plan for SMEs: a poorly secured AI project carries the same risks as a poorly protected information system, plus the specific issue of personal data processing.

The AI Act timeline after the Digital Omnibus

The European compliance timeline has recently changed. The “Digital Omnibus” regulation on AI, Regulation (EU) 2026/1744 of 8 July 2026, entered into force on 27 July 2026 and postpones several key deadlines, according to the Official Journal of the European Union.

Before investing in compliance, a Luxembourg SME should therefore check which category its use of AI falls into, since the deadlines differ significantly depending on the type of system concerned.

The Article 50 transparency obligations, which require users to be informed when content is generated or manipulated by AI, notably for certain chatbots, have applied since 2 August 2026, with an extra period until 2 December 2026 for tools already on the market before that date. This is the nearest deadline for a company already using a customer-facing chatbot or content generator, and therefore the one to check first.

For high-risk systems under Annexes I and III, the postponement gives a little more time, until the end of 2027 or 2028 depending on the case. This extra time is no reason to delay documenting your AI uses: compliance is better prepared ahead of the deadline than in the rush just before it.

Where to start: the roadmap for an SME

For a manager who wants to make progress on artificial intelligence in business without getting lost between funding schemes and regulatory obligations, the logical order comes down to four steps:

• Define the need and submit a Fit 4 AI application on MyGuichet.lu to fund 50% (SMEs) of the diagnostic with an approved consultant
• Check with the CNPD whether the project is a use case to test in the Sandkëscht sandbox before any rollout
• Apply for 70% SME Packages - AI funding for implementation once the roadmap is approved
• Document HR uses and AI-generated content to meet the transparency obligations already in force

In the space of two years, Luxembourg has built a coherent set of support measures and safeguards for artificial intelligence in business: two complementary funding schemes, an identified supervisory authority and a tool for testing before deploying. It is now up to each SME manager to make use of them before the gap with the third of companies already equipped turns into a lasting competitive disadvantage.