NVIDIA and EBU Expand Sovereign AI Cloud Framework for European Broadcasters; Adobe Accelerates Video Search with NVIDIA GPUs

Image: NVIDIA · Source

NVIDIA partners with the European Broadcasting Union to enhance sovereign AI frameworks for public media, while Adobe leverages NVIDIA GPUs to speed up video search for German producer UFA, improving editorial workflows with advanced AI and cloud technologies.

At a recent creative industry event in Berlin, hosted collaboratively by Adobe, Bertelsmann, NVIDIA and UFA, significant advancements in intelligent media technologies were unveiled. Central to these announcements was the collaboration between NVIDIA and the European Broadcasting Union (EBU) to expand a sovereign AI framework tailored for European public service broadcasters.

The EBU, comprising over 115 member organizations across more than 55 countries and reaching over a billion viewers, is focused on safeguarding editorial independence, data sovereignty, and national resilience as AI agents increasingly integrate into newsrooms and production workflows. Building on standards introduced in the prior year, the expanded initiative now includes sovereign cloud providers, open model developers, and system integrators to establish deployable services. These services will define operational requirements, reference architectures for workload deployment, data governance protocols, model usage, and end-to-end system security.

Key participating sovereign cloud providers include Microsoft, which contributes expertise from its Microsoft Sovereign Cloud offerings such as Azure Local and Foundry Local. These platforms enable broadcasters to process sensitive editorial content locally, maintain critical services during connectivity outages, and promote operational resilience. Nebius offers a European infrastructure platform for broadcasters to train, fine-tune, and run custom AI models. Meanwhile, NVIDIA's Nemotron open models provide adaptable AI foundations suitable for local languages, archives, and editorial policies.

The initiative also involves system integrators like Qvest to ensure practical and interoperable deployment of reference architectures within real-world broadcast environments. Antonio Arcidiacono, Chief Technology and Innovation Officer at EBU, emphasized that these standards furnish members with a shared foundation to bolster resilience and preserve trust through accountable AI-driven services. Delphine Ernotte, CEO of France Télévisions and EBU president, highlighted the role of sovereign cloud infrastructure and open models in enabling European media innovation while protecting cultural diversity.

In parallel, Adobe showcased its enhanced video search capabilities powered by NVIDIA accelerated computing, aimed at helping creative teams navigate and edit media content more efficiently. Utilizing NVIDIA technologies such as NVDEC for video decoding, TensorRT for AI inference optimization, and CUDA for preprocessing and embedding generation, Adobe achieved approximately 5 times faster embedding generation and up to 4 times higher throughput in video search.

Additionally, Adobe introduced GPU-accelerated media ingest workflows using NVENC and NVDEC for encoding and decoding, increasing the efficiency of managing expanding media libraries. A notable feature includes a new two-frame-per-second proxy format substantially smaller than previous versions, which reduces storage and transfer requirements while preserving sufficient visual detail for semantic search tasks.

These improvements directly benefit clients like UFA, a major German film and television producer responsible for over 3,500 broadcast hours annually, including popular shows such as "Gute Zeiten, schlechte Zeiten." Video editors can leverage Frame.io’s semantic and hybrid search capabilities to locate specific moments—such as a character’s expression—from archival footage spanning over a decade, bypassing the limitations of traditional file naming and categorization.

Adobe and NVIDIA’s ongoing partnership extends beyond video search into areas such as Adobe Substance 3D, Firefly Foundry, and 3D digital twin workflows, reinforcing their commitment to advancing media and entertainment technologies.

Together, these developments demonstrate a concerted effort to combine sovereign cloud infrastructure, open AI models, and GPU acceleration to empower European public service media and creative industry workflows securely and efficiently.