← Blog November 8, 2026

Choosing AI for everyone: what the new cabinet vision already delivers for SMEs

On 6 November, the Dutch cabinet presented its AI vision and names SMEs explicitly. We list what is settled, what is still planned, and what you can do today.

Nachtelijk polderlandschap met lichtlijnen die opstijgen naar een gloeiend netwerk

On 6 November 2026, the Netherlands got its own AI course. The Council of Ministers approved the Kabinetsvisie op Artificial Intelligence – Kiezen voor AI voor iedereen (Cabinet vision on Artificial Intelligence – Choosing AI for everyone), proposed by State Secretary Aerdts (Digital Economy and Sovereignty). It is the first AI strategy for the entire government, and SMEs (het mkb) are named directly in it. Good news, but what does it mean concretely for your business? In short: for now, not much; perhaps more later. And there is plenty you don't have to wait for.

What the cabinet decided on 6 November

The vision rests on three pillars:

Responsible development and use of AI.
Access to advanced AI technology and security.
Steering the AI transition in line with public values.

For entrepreneurs, a few points stand out. The cabinet wants to remove the barriers that currently keep SMEs from using AI. It wants to help businesses and governments apply the European AI Act in practice, including by letting them experiment with new AI applications under supervision. And from 2027, the cabinet is setting aside €120 million to participate in a European AI project, according to State of Tech led by Germany and focused on technology, infrastructure and applications.

Also noteworthy is the tone on dependency. Aerdts told the NOS that full independence from foreign tech companies is not the goal, because that is not realistic. The aim is for the Netherlands not to depend on "one country, one company, one player".

Settled, planned or still under investigation?

With policy news, it pays to make a distinction. Not everything that appears in the newspaper is settled.

Settled

The vision itself, with its three pillars, and the €120 million reserve from 2027 for the European project.

Planned

The cabinet is working on a broad approach to data centres. A plan for the AI infrastructure and computing power the Netherlands needs will follow in early 2027. A Mediapact is also being worked on, covering among other things reliable information and the use of journalistic content by AI. For now, that is still a policy intention, not a finalised agreement between government, publishers and AI companies.

Still under investigation

A Dutch AI safety institute. Some media wrote that it "is coming", but the official wording is that the cabinet is investigating the possibility. According to AIwereld, it is still not certain whether a separate institute will be created, and what tasks it would then have.

And those barriers for SMEs? The announcement and the coverage contain no concrete measures for them yet. The promise is there; the details are not yet.

The criticism: fine words, and now what?

Not everyone is enthusiastic. In an opinion piece, Dagelijkse Standaard argues that SMEs mainly get fine words and no new measures. Binnenlands Bestuur reports that the Netherlands Institute for Human Rights is sounding the alarm: it has too little capacity to properly monitor the impact of AI on fundamental rights.

On the same day, the NCTV also warned that hackers are embracing AI and that governments and businesses are not keeping up. A useful reminder: while the policy is still taking shape, the threat landscape is already changing.

It fits into a broader movement. State of Tech reports, based on Dutch IT Channel, that spending by Dutch companies on AI tools is levelling off. The purely experimental phase is over; businesses want to see demonstrable added value.

What you can do as an entrepreneur today

There is no need to wait until early 2027. A few steps that make sense today:

Check your obligations under the AI Act. Which AI do you use, for what purpose, and does any of it fall under stricter rules? The help the cabinet promises is welcome, but the rules apply regardless.
Spread your dependency. The cabinet does not want to lean on a single company; that is also a healthy principle for your business. Build your processes so that you can switch AI model or supplier without redoing everything.
Look critically at returns. Which AI applications genuinely save time or improve quality, and which have remained an experiment? Cut or improve.
Take security seriously. Given the NCTV's warning: know where your data is stored, who can access it, and what your AI tools do with it.
Keep an eye on the test environments. If experimenting under supervision becomes possible, it may be interesting for anyone who wants to build something new.

At Silvar, we see AI primarily as a tool that has to prove itself: applied in practice, with a human checking where needed, and hosted in European data centres. Are you unsure whether your current AI use is future-proof, or do you want to build an application that is not tied to a single supplier? Then we are happy to think along with you, starting with a realistic assessment of what it will require.