Paris, June 11, 2025
Jensen Huang’s keynote yesterday at NVIDIA GTC Europe and VivaTech 2025 in Paris painted a compelling vision for European AI sovereignty, complete with partnerships, billion-dollar infrastructure projects, and warnings about falling behind global competitors. But beneath the rhetoric about “sovereign AI” and “national intelligence infrastructure” lies a more complex reality that European organisations should carefully consider. 
The Sovereign AI Promise
NVIDIA’s European strategy centers on what Huang calls “sovereign AI” or the idea that every country needs autonomous AI capabilities aligned with local values and independent of foreign tech giants. The announcements are impressive: 18,000 Blackwell chips deployed with Mistral AI in France, an industrial AI cloud in Germany with 10,000 GPUs, and partnerships spanning from Barcelona’s Supercomputing Center to automotive giants BMW and Mercedes-Benz.
Huang’s message resonates with European concerns about digital sovereignty, particularly his emphasis that “the data belongs to you… your people, your country… your culture, your history.” With Europe’s AI computing capacity projected to increase tenfold over the next two years and more than 20 AI factories in development, the market opportunity is undeniable.
The Technical Reality Check
However, the sovereignty narrative requires closer examination. While NVIDIA’s hardware may be deployed within European borders, true AI sovereignty involves far more than geographic placement of servers. Modern AI systems operate through continuous learning loops, constantly updating based on global user interactions. A “sovereign” AI system running on European infrastructure still relies on models trained on internationally sourced data, using architectural patterns developed globally.
The fundamental challenge isn’t hardware location. It’s the interconnected nature of how AI systems learn and evolve. No amount of local infrastructure can change the reality that artificial intelligence is inherently global in its knowledge sources and dependencies..
What Europe Really Needs
Rather than chasing sovereignty theatre, European organisations need platforms built for transparency, governance, and genuine operational control. The real value lies not in where the servers sit, but in understanding how AI systems make decisions, who can access them, and how they evolve over time.
This is particularly relevant as European companies explore applications ranging from content localisation across multiple markets to real-time operational optimisation. Success in these areas requires distributed intelligence that can coordinate seamlessly across organisations and geographies, capabilities that transcend national boundaries by design.
Thinking 10 Years Ahead
At NVIDIA GTC Paris, where Stelia is participating alongside hundreds of other AI companies, the conversations extend beyond hardware specifications to fundamental questions about how organisations can deploy AI at scale while maintaining control and transparency. The event showcases not just the impressive technical capabilities now available, but the complex challenges of implementing AI systems that work reliably across diverse operational environments.
NVIDIA’s European expansion undoubtedly brings valuable technology and partnerships to the continent. However, European organisations would be wise to focus on platforms that deliver genuine visibility and control over their AI operations, rather than sovereignty promises that cannot address the fundamental realities of how artificial intelligence actually works.
The future of European AI competitiveness lies not in isolated national systems, but in transparent, governable platforms that enable organisations to harness intelligence without boundaries while maintaining full operational oversight.
Connect with Stelia at GTC Paris 2025
Stelia is on‑site June 10–12 at Paris Expo Porte de Versailles (Booth G32), ready to demo how Lyra can immediately operationalise NVIDIA’s compute: across edge devices, regions, regulatory zones, and into the quantum future.
To learn more about Stelia or to book a meeting during the event: visit Stelia, book a meeting or register your interest at https://lyra.stelia.ai
