TensorX raises €8M to build sovereign AI infrastructure for Europe

The launch of a dedicated European inference layer addresses a critical bottleneck for highly regulated sectors like finance and healthcare that require strict data residency. By utilizing localized Nvidia hardware, TensorX provides a compliant alternative to US-based cloud giants, mitigating business continuity risks associated with model availability across borders. This move signals a shift toward specialized 'Neo Clouds' that prioritize jurisdictional control over raw compute scale. As the EU AI Act takes effect, watch for the planned €100 million infrastructure expansion to see if TensorX can maintain performance parity while meeting stringent sovereignty requirements.

TensorX enters a rapidly crowding field of European sovereign AI infrastructure providers, each racing to capture regulated enterprise workloads before the EU AI Act's enforcement deadlines tighten. In May 2026, Nebius Group, the Amsterdam-based AI cloud spun out of Yandex, raised $700 million in a Series D round to expand GPU clusters across Europe, positioning itself as a direct competitor to US hyperscalers for sovereignty-sensitive workloads. Nebius operates data centers in Finland, the Netherlands, and France, and its funding round valued the company at approximately $10 billion, underscoring investor appetite for European AI compute alternatives. TensorX's €8 million seed round is modest by comparison, but its focus on inference-only deployments for regulated verticals differentiates it from the broader training-and-inference platforms that larger players are building.

The regulatory backdrop driving demand for sovereign AI infrastructure is accelerating. The EU AI Act's high-risk classification requirements began phasing in during 2025, and the European Commission published updated guidance in March 2026 clarifying that organizations deploying high-risk AI systems must demonstrate full data residency and auditability within EU jurisdiction. That guidance effectively mandates that financial services firms, healthcare providers, and public-sector bodies either build or procure inference infrastructure on European soil. TensorX's Dublin and Helsinki GPU clusters, built on Nvidia Blackwell architecture, are designed specifically to satisfy those compliance requirements. Meanwhile, Nvidia announced in June 2026 that it had certified 14 European data center operators as sovereign AI cloud partners, a program that provides co-marketing, optimized software stacks, and reference architectures for GDPR-compliant deployments.

On the technical side, inference performance on Nvidia Blackwell GPUs has become a key differentiator among sovereign AI providers. MLPerf Inference v5.1 results published in April 2026 showed that B300-based systems achieved up to 2.3x throughput improvement over H100 clusters on Llama 3 70B workloads, a benchmark directly relevant to the open-source model deployments TensorX targets. TensorX co-founder Tim Grant has indicated the company plans to scale toward a €100 million infrastructure buildout, which would place it in the same tier as mid-size European GPU cloud operators. The competitive pressure from both well-funded startups like Nebius and hyperscaler sovereign offerings from AWS and Azure means TensorX must demonstrate measurable latency and cost advantages within its regulated-industry niche to justify its expansion timeline.