SOTEU 2026: Europe wants AI in the factory, not just on the screen

On 16 September 2026, before the European Parliament in Strasbourg, Ursula von der Leyen delivered her State of the Union address. Of the two “tipping points” of our time she singled out — climate and artificial intelligence — AI was presented as a foundational layer of Europe’s economy and security.

For the leadership of a European SME or mid-sized company, this is not an abstract speech. It is a useful frame: Europe stands by its guardrails (the AI Act), wants to stay in the race on computing capacity and, above all, to shift the battle from the frontier model to applied value in the real economy.

What the speech actually says

The Commission President drew a clear line: make the most of AI for society, quality of life and productivity, while addressing the risks — otherwise those opportunities will stay locked.

She recalled that the AI Act is a crucial piece in setting guardrails, and that Europe intends to shape international efforts (model evaluation, verification, early warnings, AI safety) with partners such as Canada and the United Kingdom. She also announced that she wants to bring together the leading frontier labs to discuss a more responsible pace for self-improving models.

At the same time, Europe will need to massively scale up its computing power and devise new public-private financing arrangements for its most promising start-ups and companies.

The value is not only in the model

The most operational passage for businesses is not about the race for the biggest model. Ursula von der Leyen put it this way: Europe does not need to be the one that invents the frontier technology to get the most value out of it — as was once the case with the printing press.

The challenge, she said, is to move AI from the screen into the real economy: the workshop, the care ward, precision farming, energy grids, autonomous driving, defence.

And Europe has a concrete asset: world-class industries built on high-quality data. That data, she stressed, is a European treasure for training tailor-made industrial models.

Five priority sectors — initiatives in November

The Commission is now focusing its effort on five high-value sectors, where industrial AI has the most potential and where Europe has the data:

1. health;
2. transport;
3. agri-food;
4. advanced manufacturing;
5. defence and space.

In November, the Commission will announce “game-changing” initiatives in these sectors and will invite Member States, parliamentarians, industry and entrepreneurs to join the project.

The stated approach: more value, more responsibility, at lower cost and with less energy use — and above all AI that preserves human agency, with humans in control.

The health example: assist, not replace

The speech illustrated the point with breast cancer screening. Ursula von der Leyen recalled that mammography screening reduces mortality by about 30 to 40% among the women who take part, and that AI can help detect more cancers, earlier. Then she was clear: yes to a doctor with instant access to the data they need thanks to AI; no to a robot announcing the diagnosis on its own. AI must strengthen the doctor, not replace them.

This frame — a decision-support tool, clear human responsibility — is exactly the one many European management teams are already trying to set in their own processes (customer support, quality, maintenance, finance).

Three concrete steps for European leadership teams

1. Map where AI already touches the real world. Support functions (emails, summaries) or the core business (production, quality, logistics)? The European speech explicitly pushes towards the latter.

2. Treat data as an industrial asset. If Europe’s advantage lies in the quality of its sector data, the priority is not “which LLM?” but “which data, which governance, which process?”.

3. Keep humans in the loop — and document it. Transparency, responsibilities, named use cases: an approach in line with the spirit of the AI Act, several of whose transparency obligations already apply, and with the SOTEU’s “human agency” line.

What AZINOVE SAS takes from it for its European clients

At AZINOVE SAS, we design and integrate useful AI for European management teams: software development, consulting, putting systems into production. SOTEU 2026 confirms what we see on the ground: the companies that move forward do not collect demos — they anchor AI in a process, with clean data and clear responsibility.

The right order remains:

1. use case (value and risk);
2. data and process (where AI really takes hold);
3. tool (SaaS, custom build or hybrid);
4. measurement (before / after) and governance (humans in control).

November will bring sector announcements from the Commission. Until then, leadership teams do not have to wait: they can already pick a process, name a key performance indicator (KPI) and take AI beyond the screen.

Sources (in French)

1. Commission européenne — 2026 State of the Union Address by President von der Leyen (16 septembre 2026, Strasbourg) — texte « check against delivery », Representation in Cyprus : https://cyprus.representation.ec.europa.eu/news/2026-state-union-address-president-von-der-leyen-2026-09-16_en
2. Service audiovisuel de la Commission — débat SOTEU 2026, déclaration sur l’IA (transcription) : https://audiovisual.ec.europa.eu/en/media/video/I-293059
3. La Dépêche — couverture du discours (16 sept. 2026) : https://www.ladepeche.fr/2026/09/16/ursula-von-der-leyen-devoile-la-strategie-ia-de-leurope-priorite-a-lutilisation-responsable-et-a-la-valorisation-des-donnees-europeennes-13554261.php