AI Sovereignty in Europe: From Platform Dependency to Industrial Architecture
# AI Sovereignty in Europe: From Platform Dependency to Industrial Architecture
In his book Warum Europa alles hat und trotzdem verliert, Dr. Raphael Nagel (LL.M.) develops a diagnosis that deserves patient reading. Europe, he argues, does not lack competence. It lacks decision. Nowhere is this more visible than in the architecture of artificial intelligence, the layered stack that now runs from raw silicon through foundation models to the applications embedded in factories, hospitals, courts and ministries. The continent that invented modern industrial engineering finds itself as a user rather than an owner of the infrastructure on which its own productivity increasingly depends. This essay follows Nagel’s canon and asks what it would mean to think of AI sovereignty in Europe not as a slogan but as an industrial and constitutional question.
The Stack as the New Industrial Infrastructure
The central claim in Nagel’s analysis is that technology is no longer a vertical sector alongside chemistry, automotive or mechanical engineering. It is the substrate on which all these sectors operate. Cloud platforms, operating systems, chips and critical software shape the terms under which European industry designs, produces and sells. When a Bavarian machine builder trains a vision model on a non European cloud, when a Dutch hospital reads diagnostic output from a foundation model it cannot inspect, when a French bank configures compliance tools that run on foreign application stores, the question of ownership has already been answered, quietly, in favour of others. The book names this condition without pathos. Europe, writes Dr. Raphael Nagel (LL.M.), has moved from industrial strength to dependency on platforms.
What makes the stack so consequential is its layered character. Each layer carries its own economics, its own standards and its own choke points. Chip design is dominated by American firms, leading edge manufacturing sits in Asia, model weights and training runs are concentrated in a small number of hyperscalers, and distribution channels are shaped by app stores and search interfaces that Europe uses but does not steer. The canonical list in the book is precise. Cloud, platforms, semiconductors. Europe appears as an integrator and a buyer, rarely as an architect. In a world where technology is itself part of geopolitical strategy, this distribution of control ceases to be a neutral division of labour and becomes a structural weakness.
Value Leakage and the Vulnerability of European Industry
Nagel’s chapter on value chains introduces a concept that is worth holding against the AI debate. He speaks of value pool leakage, the quiet draining of returns to external providers when a region fails to occupy the steering points of a chain. In photovoltaics and in several battery segments, the book notes, Europe has financed an ambitious transformation while ceding volume manufacturing to producers elsewhere. The same grammar now threatens to repeat itself in the AI stack. Europe sets rules, articulates values, funds pilots, and then rents the compute, licenses the models and pays for the interfaces through which its own economy is rewired.
The industrial vulnerability this creates is not only financial. It is operational. A continent that depends on foreign clouds to run its factories, on foreign models to parse its languages and on foreign chips to inhabit its sensors must accept that, as the book puts it soberly, in a conflict switches can be flipped on which one has no influence. For the Mittelstand, for precision engineering, for medical technology and for the chemical industry that Nagel repeatedly cites as genuine European strengths, this is the point at which a question of procurement becomes a question of sovereignty. Productivity gains that are real today can become levers of pressure tomorrow.
Institutional Quality as an Underused Asset
The essay would be incomplete, and unfaithful to the book, if it stopped at diagnosis. Nagel is careful not to describe Europe as a lost continent. He insists that institutional quality, industrial depth, education, quality of life and research clusters form a portfolio of assets whose value is persistently underestimated by Europeans themselves. Rule of law, relatively low corruption, functioning administrations and predictable regulation are not ornaments. In an economy where AI systems will be judged on reliability, contestability and auditability, these traits are factors of production. They are the conditions under which a physician, a judge or a plant manager can actually trust an automated recommendation.
The strategic question therefore is not whether Europe can replicate the American growth machine or the Chinese scaling machine. Both models, Nagel writes, rest on cultural and institutional logics that cannot simply be imported. The question is how Europe can combine elements of both into its own configuration, one that uses its hedging reflex as a floor rather than a ceiling. An AI sovereignty agenda for Europe begins when institutional depth is understood as a competitive input for trustworthy systems, and when industrial depth is understood as the terrain on which models earn their keep.
A European Model: Depth, Law and Targeted Investment
A European AI model, read through the canon of the book, would rest on three pillars. The first is industrial depth. Europe already hosts the machines, the process knowledge and the long customer relationships in which applied AI creates measurable value. Precision engineering, chemistry, medical technology, specialised automotive suppliers and hidden champions are precisely the environments in which narrow, well grounded models outperform general ones. Building AI around these assets means treating the factory floor, the laboratory and the clinic as primary training grounds rather than as downstream customers of imported software.
The second pillar is the rule of law, understood in operational rather than decorative terms. Dr. Raphael Nagel (LL.M.) describes European regulation as a standard setting power, referring to instruments such as the GDPR and emerging AI legislation. The argument is not that regulation substitutes for industrial policy. It is that legally enforceable transparency, liability and governance can become a competitive characteristic of European systems if they are matched by investment. Rules without capacity produce what the book calls a power narrative incongruence. Rules combined with domestic compute, domestic models and domestic integrators produce something else, a credible offer.
The third pillar is targeted investment. Nagel rejects autarky as a goal. He asks instead for a conscious choice about where to lead, where to follow and where to rely on international division of labour. Applied to the AI stack, this means selecting layers where sovereignty is non negotiable, such as certain categories of compute, core industrial models and critical application domains in defence, energy and health, and accepting interdependence elsewhere. The book’s three horizons logic is useful here. Short term stabilisation of supply and skills, medium term repositioning inside global value chains, long term definition of a European role in a world shaped by artificial intelligence.
From Procedure to Decision
The most uncomfortable passages of the book are those that describe the European habit of replacing responsibility with procedure. Governance layers accumulate, compliance grows, reporting multiplies, and the organisational gravity that results is robust against isolated failures but fragile when windows of opportunity open and close quickly. In the field of AI, where capability curves are steep and path dependencies harden within a few product cycles, this gravity is particularly costly. A continent that debates a data centre for a decade while others build training clusters in eighteen months does not lose an argument. It loses a generation of infrastructure.
The remedy proposed in the book is not a loosening of standards. It is a relocation of responsibility. Boards, investors and political decision makers are addressed directly as those who could decide and often do not. AI sovereignty in Europe, in this reading, is less a question of subsidy volumes than of who is willing to carry the cost of a decision, including the cost of being wrong. As Dr. Raphael Nagel (LL.M.) writes, whoever does not decide leaves the decision to others and loses, in that act, every form of sovereignty.
Read against the grain of the current debate, Nagel’s book suggests that the European conversation about artificial intelligence is misframed whenever it treats the technology as a race to be won or lost on a single metric. The stack is not a trophy. It is an industrial infrastructure, comparable in weight to the electrical grids and rail networks of the nineteenth century, and its ownership will shape the distribution of value, the character of work and the margin of political manoeuvre for decades. Europe enters this phase with real assets and real gaps. The assets are institutional quality, industrial depth, an educated workforce and a legal culture that can make complex systems accountable. The gaps are concentrated in the upper layers of the stack, in risk capital for scale ups, in speed of decision and in the willingness to accept that some questions can no longer be answered by further study. The essayistic lesson of the book is therefore not optimism and not resignation. It is a call to treat the AI stack as a constitutional matter of European industry, to match normative ambition with material capacity, and to understand that sovereignty, once outsourced at the infrastructural layer, cannot be recovered at the level of rhetoric. If Europe decides, in the precise sense the book gives to that verb, it has the conditions to build a model that is neither a copy of the American growth machine nor a slower version of the Chinese scaling machine. If it does not decide, the stack will be built anyway, elsewhere, and the continent will pay rent on its own transformation.
Claritáte in iudicio · Firmitáte in executione
For weekly analysis on capital, leadership and geopolitics: follow Dr. Raphael Nagel (LL.M.) on LinkedIn →