While news in the U.S. and China this summer has been dominated by one AI model after another and massive investments in computing power and AI startups, Europe has been busy implementing and revising AI regulations. For example, as of August 2, the European AI Office can fine providers of the most powerful AI models, and anyone marketing AI-generated content must label it with a logo or watermark.
Meanwhile, criticism of the European AI Regulation is growing, as strict regulation of AI is seen as an obstacle to European competitiveness. So, on July 27, another European regulation went into effect: the Digital Omnibus. It postpones the requirements for high-risk AI until December 2027, and for AI in products such as medical devices until August 2028. In the same week, Europe thus began both enforcing and postponing the regulations.
This erratic course is disastrous for investment. Moreover, the notion that regulation is, by definition, bad for innovation is misleading. Europe itself has proven that well-designed regulation can actually stimulate innovation. In the 1980s, Brussels mandated a single mobile standard—GSM—and reserved spectrum for it. This gave rise to a single market that spawned companies like Nokia and Ericsson. That is regulation for the sake of innovation: a rule that creates a market. The AI Regulation does not do this: it was written to prevent harm. That is a legitimate and valuable goal, but it does not contribute to European competitiveness. The real problem is that, by purchasing digital products from American suppliers for decades, Europe has come to confuse regulation with industrial policy.
Europe is beginning to realize this and is seeking ways to strengthen its sovereign position in the global AI race when it comes to access to AI, control over AI, and the value it creates. Sovereignty does not require Europe to control the entire AI chain itself, but rather to have its own position in strategically crucial areas, including AI infrastructure and state-of-the-art technology. This also includes resilience—on the part of citizens, governments, institutions, and businesses—against unsafe AI, the undermining of institutions, and economic dependence. The combination of responsible, sovereign, and resilient AI is called AI Fair Tech. Europe holds the cards for AI Fair Tech: high-quality data, top talent, strong institutions, sufficient capital, and the second-largest market in the world. It must play those cards in a coordinated manner.
A strategically relevant position in AI requires a coherent European AI industrial policy centered on European norms and values. First and foremost, the AI Regulation must evolve into a regulatory framework that promotes desirable AI—and, by extension, innovation and investment. But regulation alone will not be enough for Europe. Regulation must be carefully aligned with the two other instruments that are intended to support the economic and societal AI transition: AI innovation policy and public AI procurement policy.
In June of this year, Jim Hagemann Snabe was appointed as the EU Special Envoy for Industrial AI; his term runs through March 2027. AI industry policy cannot be addressed with just a temporary envoy. We therefore advocate for the establishment, at both the European and member state levels, of public AI industry policy centers where coherent AI industry policies are developed regarding regulation and oversight, innovation policy, and public procurement. These centers must be assigned a legally defined role vis-à-vis governments. In September, Von der Leyen will deliver her State of the Union address, and negotiations on the European multiannual budget will begin. That is the moment to transform the envoy’s mandate into a European center for AI industry policy, with its own independent budget.
There is broad consensus that Europe must invest in groundbreaking AI innovation. That is why we advocate establishing, in addition to the EU network of AI factories, a European network of public-private AI frontier centers where top talent works at the cutting edge of the technology. These centers are small-scale, equipped with all the necessary conditions to attract top talent—including special access to AI infrastructure—and feature a dynamic workforce where talent moves through the system quickly, ensuring a continuous flow of new ideas and innovations. They thus serve as incubators for startups and technology transfer.
The Netherlands doesn’t have to wait for that. Various programs are already underway, and there are plenty of proposals from the field; the Wennink report contains an exhaustive list. Decisive action is now required. In this regard, we advocate taking another critical look at the many proposals for new institutes, labs, and agencies. Some of these activities can be better achieved through new programs rather than through new organizations. New organizations introduce complexity, and they all compete for the same pool of scarce top AI talent. We see room for only two complementary, public-private centers: an AI frontier center focused on state-of-the-art
AI development and application, where top talent in the field of AI technology works; And an AI industry policy center that advises the government on regulation, oversight, innovation, and procurement. This is where the top talent in AI industry policy is based. We call on the government and the business community to allocate the necessary resources to establish these centers in the near term.
A more detailed elaboration of this argument can be found in our position paper, “The European AI Industry”: https://aic4nl.nl/wp-content/uploads/2026/09/Position-paper-Europese-AI-industrie.pdf
