Brussels Has Picked Manufacturing for Its Next AI Push. The Entry Ticket Is Your Data
On 16 September, Ursula von der Leyen used her State of the Union address to name five sectors where, in her words, industrial AI has the biggest potential and Europe has the data: health, transport, agri-food, advanced manufacturing, and defence and space. She promised game-changing initiatives across these sectors in November and invited member states, MEPs, industry and entrepreneurs to take part.
Political speeches are easy to skim past. This one deserves a closer read from anyone running production in Europe, because the argument behind it is unusually concrete, and because the November announcements are likely to turn into calls, consortia and funding that manufacturers can actually apply for.
The argument: Europe does not need to win the model race
The core of the speech was a reframing. Von der Leyen argued that the next leg of the AI race is not only about frontier models, and used the printing press as an analogy: invented by a German, but mastered and applied across Western Europe. Her conclusion was that Europe does not need to develop the frontier technology to draw the greatest value from it.
Where she located that value is the part worth noticing: moving AI off the screen and into the real economy, including the factory floor. Europe, she said, has world-class industries sitting on the highest-quality data, and those data can power tailor-made industrial AI models.
This is not new policy so much as a sharper version of the direction already set. The Apply AI strategy, a €1 billion initiative covering ten strategic sectors, already foresaw agentic AI being deployed to optimise production lines and logistics from the fourth quarter of 2026. What changed on 16 September is the emphasis. Manufacturing is now explicitly one of five priority sectors, and data is named as the asset that decides who benefits.
The catch in “we have the data”
At the level of the continent, the claim is true. European industry generates enormous volumes of machine, process and quality data. At the level of an individual company, the picture is usually different. The data exists, but it sits in a controller nobody exports from, in a quality system that produces PDFs, in an MES from 2011, and in the heads of three people. Commentators were quick to point out that data availability remains a major roadblock for European innovators, and they are right.
That matters for what comes in November. EU programmes rarely hand out money to companies for good intentions. They fund concrete use cases, usually in consortia, and they favour participants who can show that the data a project depends on actually exists and can be used. The manufacturers who will benefit from this push are not necessarily the largest. They are the ones who arrive with a defined problem and data in a usable state.
What to do in the next eight weeks
Nobody knows yet what exactly will be announced, and “game-changing” is political language. But the groundwork is the same whatever form the initiatives take, and it pays off even if you never apply for anything.
Start by deciding what problem you would bring. Not “AI in production”, but one specific thing: scrap on a particular line, changeover times, quoting for complex parts, maintenance on a critical machine. Then check the data behind it. What exists, in what form, going back how far, and who controls it – you or the supplier of the machine that generates it. Since last year the EU Data Act has given users rights to the data their connected machines produce, which is worth knowing if the answer is “the machine builder”.
Finally, talk to your industry association or regional cluster now rather than in November. Consortia for EU calls tend to form quickly and around organisations that are already in the conversation. A mid-size manufacturer with a clear use case and usable data is exactly the partner these consortia look for, and rarely the one that gets found last-minute.
The next phase of Europe’s AI strategy is aimed squarely at factory floors like yours. Whether you can take part depends less on how much you know about AI and more on whether your data is ready when the doors open.
“Content generated using AI”