- Europe aims to build AI sovereignty but remains heavily dependent on Amazon, Microsoft, and Google for cloud infrastructure.
- DeepL once promoted data processing in Germany and Iceland, but since April 2026, it has partnered with AWS to expand its services globally.
- Multinational enterprises require low latency, global deployment capabilities, and distribution channels that European cloud providers have yet to meet.
- The article argues that DeepL’s decision reflects market pressure rather than an abandonment of technological sovereignty goals.
- Europe’s AI policy focuses on building model training capacity, but actual demand lies in AI inference to serve users.
- Proprietary models such as GPT, Claude, and Gemini primarily run on the infrastructure of US hyperscalers.
- European cloud providers mostly operate open-source models, often outdated or scaled-down versions due to GPU cost constraints.
- Nvidia supports new AI cloud providers in Europe, but many units still have to serve the needs of US tech corporations.
- The report suggests that AI demand is increasingly tied to cloud ecosystems and proprietary services, making it difficult for businesses to switch providers.
- The author concludes that increasing investment in AI or cloud separately is not enough for Europe to achieve AI sovereignty without addressing cloud market concentration.
📌 The close link between AI sovereignty and the global cloud market. The case of DeepL shows that even successful European AI companies find it difficult to avoid dependence on AWS when they need scale, performance, and corporate clients. The author argues that merely expanding model training capacity or investing in infrastructure is not enough to reduce dependence, as AI demand is now tightly bound to proprietary models and cloud ecosystems controlled by US hyperscalers.
