EU AI Strategy Report Warns of Dependency, Proposes Controversial Trade-Off

A new report titled ‘Transformative AI Strategy for Europe’ has sparked debate over the continent’s approach to artificial intelligence. Released last month by a coalition of economists, politicians, and AI experts, the report diagnoses Europe’s deep dependency on overseas providers of essential AI technologies. It introduces the concept of ‘Transformational AI,’ which is gaining currency among policy elites.

The report carries weight due to its high-profile signatories, including Margaret Vestager, the former vice president of the European Commission. Vestager issued a stark warning: “We cannot allow Europe to become a playground for foreign powers. Our citizens are already bearing the economic and social consequences of decisions taken elsewhere. AI will accelerate this development, transforming our personal lives and relations, learning and jobs at an unprecedented pace.”

European Central Bank chair Christine Lagarde, though not connected to the report, has similarly highlighted the risks. She noted that a withdrawal of access to AI models or a change in their terms would reach every sector at once, representing leverage no trade partner has ever held over Europe. Such leverage, she said, could be used in negotiations on tariffs or digital taxes.

Two Paths Forward

The report outlines two possible approaches for Europe. The first is to invest heavily in building European frontier models that could compete on equal footing in the transformational AI economy. However, the report’s signatories dismiss this path as too costly and too high risk. Instead, they strongly recommend a second approach: trade power and land for access to frontier models.

Under this proposed strategy, Europe would fast-track the build-out of Big Tech infrastructure on European soil. In exchange, Europe would receive “guaranteed” preferential access to frontier models. Critics argue this amounts to capitulation to mega corporations under the control of foreign regimes. One author compared it to Native Americans trading land for trinkets.

Critique of the Recommended Approach

The author of the critique contends that it is naïve to expect lasting privileged access to frontier models in return for power and land in Europe. Big Tech already has this access, and Europe holds a weak hand in trying to extract conditions for inward investment. Corporations and nation states are already leveraging the economic and political power of leading frontier models, with export restrictions on the latest models already in place.

Moreover, mark-ups of 20 to 50 percent are imposed to run models within defined jurisdictional boundaries, such as EU clouds. The author points to OpenAI’s use of an internal tool, described as “significantly more capable” than its most recent public release, to solve part of the Navier-Stokes equation. This suggests that companies keep their best models for internal use, making it inconceivable that Europe could negotiate access to the true frontier.

A Sovereign Alternative

The author argues that Europe needs sovereign capacity and models for transformational AI. Europe does not have to accept the Big Tech narrative on what or where the frontier is. Deciding how AI is deployed to support European values is within Europe’s control and responsibility. Regulation enforces shared European values, including product safety, fairness, and personal freedom.

Europe may never catch up to US or Chinese frontier large language models, but the author questions whether that is the right race to run. The first question should be: what do we really want AI for? Europe can work on its own large but focused AI models to leapfrog and create an AI transformation conforming to European ideals and values.

Action is needed urgently, but the first step cannot be negotiating away assets such as land, power, data, and values. The author urges Europe to stop giving new workloads to overseas AI and cloud giants by default. European alternatives exist for the majority of cloud services, and European and open weight models can run with European providers. Gaps can be filled with bespoke European solutions. This would create reliable demand for homegrown AI models and compute capacity, allowing sovereign models to train on sovereign data while depriving overseas models of European resources. Nudging public procurement towards European players would grow the market and create reliable demand for venture capital and corporate funds. The paradigm must be flipped so that the default action is to consider a European solution first.