- The European Commission is being criticized for its plan to spend €20 billion (~$21.6 billion) to build massive AI computing centers.
- President Ursula von der Leyen proposed building 4–5 “gigafactories,” each using approximately 100,000 GPUs to train AI.
- The plan aims to compete with major projects like the $500 billion Stargate AI data center.
- However, many experts argue that Europe does not have enough AI companies to utilize such vast resources.
- Currently, the EU has almost only Mistral AI as a prominent company, but it is building its own infrastructure rather than waiting for the EU project.
- The EU previously planned 19 AI supercomputers across 16 countries, but the effectiveness remains unclear.
- There are 76 construction proposals across 60 locations, but the implementation process has been delayed multiple times.
- Lawmakers are questioning the “business case”: who will use them once they are built?
- Additionally, the project risks making the EU more dependent on Nvidia chips.
- Some suggest the EU should focus on its industrial strengths instead of racing to build large AI models like the US.
📌 Conclusion: The EU’s €20 billion plan to build large-scale AI centers is facing significant doubt regarding efficiency and actual demand. While the US invests up to $500 billion and China accelerates sharply, Europe lacks both AI enterprises and a clear strategy. Without adjustment, the project risks becoming a “cathedral in the desert”—costly but offering no real competitive advantage.
