Johanna Ballesteros, a Research Associate at the Oxford Internet Institute, has published an argument that Europe’s digital sovereignty debate fails to distinguish between two separate objectives: securing the technology it already depends on, and building capacity to shape the technology it will depend on in the future.
Ballesteros defines “present sovereignty” as resilience in existing systems, and “future sovereignty” as developing capabilities in emerging technological domains before competitive positions are locked in. Her central claim is that an excessive focus on closing the current gap — what she calls “presentism” — risks spending resources catching up on a frontier that has already moved.
Her analysis points to several data points to characterize the current position. AWS, Microsoft Azure, and Google Cloud account for approximately 65 to 70 percent of European cloud infrastructure, according to Synergy Research Group, as cited in the piece. The European Investment Bank reports EU R&D spending at roughly 2.2 percent of GDP, compared to 3.5 percent in the United States and 4.9 percent in South Korea. The European Patent Office data cited by Ballesteros shows Europe’s share of global AI patents declined between 2018 and 2023 relative to both the US and China.
Ballesteros draws on Mariana Mazzucato’s work through the UCL Institute for Innovation and Public Purpose to argue for mission-oriented public investment that tolerates high failure rates — the model she argues produced GPS, touchscreens, and the internet. She also references Carlota Perez’s work on technological paradigms to argue that windows of opportunity in new sectors are real but narrow, and that Europe has missed several.
Her policy proposal centers on what she calls “strategic necessity” as a science policy design principle. She suggests funding conditions could require research teams to demonstrate how their systems would function without access to key external inputs — such as a specified model, chip supply, or compute source — before funding is approved. She cites SPRIND’s Next Frontier AI Initiative as one existing example of this orientation.
Ballesteros acknowledges a counterargument: “Constraints do not automatically produce innovation: in many cases, restrictions simply force substitution with worse alternatives.” She names DeepSeek as a case where tighter compute access led to efficiency-driven training innovations, and the space race as a historical case where geopolitical constraint translated into compounding research capacity, while leaving open whether the pattern generalizes.
The piece identifies quantum technologies, scientific AI, and fusion energy as domains where Ballesteros argues Europe could pursue a leapfrog strategy, on the basis that “technological trajectories and institutional arrangements are still less settled” in those areas than in cloud or semiconductors. It does not provide independent sourcing for the claim that those trajectories remain open. Germany’s High-Tech Agenda is cited as a practical example already moving in this direction.
The post links to an interactive economic model at sovereignty-model-production.up.railway.app that Ballesteros describes as a companion tool for exploring the argument’s underlying logic.