NVIDIA Expands Sovereign AI Framework for EBU and Accelerates Adobe Frame.io Video Search

• EBU and NVIDIA are expanding their sovereign standards for European public service media to include cloud providers, open model developers, and systems integrators.
• Adobe Frame.io uses NVIDIA accelerated computing to achieve 5x faster embedding generation, up to 4x higher throughput, and up to 5x lower latency for video search.
• The expanded EBU initiative defines requirements for workload placement, data governance, and model selection to protect editorial independence and data sovereignty.
• Adobe's GPU-accelerated pipeline utilizes NVIDIA NVDEC, TensorRT, and CUDA, enabling semantic search for specific video moments by concept, action, or setting.

Sovereign Cloud and Model Partners

• Microsoft contributes Azure Local and Foundry Local to allow local processing of sensitive editorial material and maintain critical services during connectivity outages.
• Nebius provides a platform for training and fine-tuning models on European infrastructure, while NVIDIA Nemotron open models offer a customizable foundation for domain-specific applications.
• Qvest and other system integrators provide consulting to ensure reference architectures are practical and interoperable in real-world broadcast environments.

Media Production Applications

• UFA, a German film and TV producer, uses the accelerated search to find specific moments in over 3,500 hours of annual programming, including archives dating back 15 years.
• A new two-frame-per-second proxy is 80% smaller than the 360p version it replaces, reducing storage and transfer costs while maintaining visual detail for search.