America can switch off AI. Europe must switch gears before it's too late
The article argues that the United States holds significant control over global AI infrastructure, warning that Europe risks falling behind and becoming dependent. It urges Europe to accelerate its own AI development, investment, and regulatory approach to avoid being left vulnerable to US dominance in the sector.
Background
- The article argues that the US controls the essential infrastructure of AI — cloud computing (AWS, Azure, GCP), advanced chips (Nvidia GPUs), and frontier models (OpenAI, Google, Meta) — and could restrict Europe's access through export controls or sanctions, effectively "switching off" AI for the continent.
- Europe has few sovereign alternatives at any layer of the AI stack: it relies on US cloud platforms for compute, US chips for training, and US or Chinese models for deployment.
- This is the latest chapter of a long-running EU "digital sovereignty" debate, which previously centred on GDPR, data localisation, and cloud certification (Gaia-X). AI has raised the stakes because the hardware and capital needed are far larger.
- Key EU responses include the AI Act (the world's first comprehensive AI regulation), plans for shared "AI factories" (supercomputing centres), and proposals for a "CERN for AI" to pool resources.
- The warning parallels arguments made during the US CHIPS Act (2022) and EU Chips Act (2023): Europe risks becoming a consumer of foreign technology rather than a builder.