Without its own AI backbone, Europe will be a powerless rentier
Without sovereign infrastructure, even the continent’s most talented researchers are left competing on borrowed machines, governed by foreign jurisdictions and exposed to export restrictions
Artificial intelligence is no longer an emerging technology – it is the foundation of modern power. Control over AI infrastructure will determine which economies lead and which follow. For Europe, this is no longer a distant concern. It is a strategic crisis unfolding in real time.
At the core of AI lies computing power – the digital fuel that drives every breakthrough model and application. Today, that power is largely foreign-owned. US companies hold roughly three-quarters of global AI supercomputing capacity and control most of Europe’s own cloud infrastructure. China is racing ahead with more than a million high-performance processors deployed this year alone. Europe, by comparison, accounts for less than 5% of global AI compute.
This imbalance is not merely technical – it is geopolitical. Nations that depend on others for their critical digital infrastructure surrender more than efficiency; they surrender autonomy. Europe has already learned this lesson in energy. It cannot afford to repeat it in AI.
The European Union prides itself on regulation and ethics, but leadership in AI will not come from rules alone. It will come from the ability to compute. Without sovereign infrastructure, even the continent’s most talented researchers are left competing on borrowed machines, governed by foreign jurisdictions and exposed to export restrictions. This is how digital dependence becomes structural.
What Europe needs is scale – and speed. Fragmented projects and scattered R&D initiatives cannot match the industrial strategies now transforming the United States and China. The answer is to build AI gigafactories: large-scale, cloud-based infrastructures operated by European companies that deliver computing power as a service. Governments would not need to own these facilities. Their role should be to act as anchor customers, providing the long-term contracts that attract private capital and unlock the scale required to compete globally.
This model has already worked elsewhere. Just as battery gigafactories made Europe a player in the green transition, AI gigafactories can do the same for the digital one. They would provide secure, affordable computing for startups, research institutions, and public bodies, while keeping profits, innovation, and jobs within the EU. Sovereign compute capacity is not protectionism – it is the price of resilience.
Time, however, is not on Europe’s side. The United States and China are expanding at breakneck pace. xAI’s Colossus already runs 200,000 GPUs, and China’s buildout surpasses anything seen before. Europe’s target of 500,000 GPUs by 2030 is not leadership – it is survival. If this infrastructure is not operational by 2027, Europe will have lost another decade.
Artificial intelligence will underpin everything from defence to healthcare, from climate modelling to education. Allowing others to control the hardware behind it means outsourcing not just data, but destiny. Europe’s challenge is not to catch up in innovation – it is to reclaim its capacity to decide its own technological future.
The window is closing fast. Europe can continue to consume technology built elsewhere – or it can build the backbone of its own AI era. The choice will define whether Europe shapes the future, or merely rents it.
Agnieszka Łukaszczyk is a specialist in space policy at Andart Global, a consultancy firm. She previoulsy worked at the European Commission in roles related to space policy and research.
