AI Affairs, home

Monday 19 October 2026

Corporate

Nvidia and EBU expand broadcaster AI framework with Microsoft and Nebius

The initiative will set deployment requirements for AI agents used by public broadcasters, with Qvest providing systems integration and Nvidia offering customisable open models.

The glass-fronted European Broadcasting Union building and entrance in Le Grand-Saconnex.
Photo: *_*, CC BY 2.0, via Flickr (cropped)

Nvidia and the European Broadcasting Union expanded their sovereign AI framework on 19 October 2026, adding Microsoft, Nebius and Qvest to an initiative for broadcasters serving more than 1 billion people, Nvidia said. The companies will work on deployment requirements for AI agents and on ways for public media organisations to retain control of their data, infrastructure, models and security.

Key points

  • Microsoft will bring Azure Local and Foundry Local to the framework for local processing of sensitive editorial material.
  • Nebius offers European infrastructure on which broadcasters can train, fine-tune and run their own models.
  • Qvest will provide media consulting and systems integration as the group develops reference architectures.
  • The work extends a partnership announced in June 2025 around sovereign AI and hybrid cloud systems.

EBU extends its 2025 broadcaster framework

The expansion moves the EBU and Nvidia’s work from common standards towards requirements for services broadcasters can deploy. Nvidia says the group will define where AI workloads run, how data is governed, which models underpin services and how systems remain secure throughout their use. Cloud providers, open model developers and systems integrators are joining that work.

The two organisations announced their initial partnership on 11 June 2025 at Nvidia GTC Paris during VivaTech, Crypto Briefing reported. That arrangement covered sovereign AI and hybrid cloud frameworks for European public service media. It brought together the EBU’s Dynamic Media Facility and Media eXchange Layer architectures with Nvidia’s GB200 Grace Blackwell Superchip, Holoscan for Media software and DGX Cloud. Nvidia was also to offer training through its Deep Learning Institute.

The member network provides the scale for the framework, although the two accounts give different counts. Nvidia puts the EBU at more than 115 member organisations in over 55 countries, reaching more than 1 billion people. Crypto Briefing puts its membership at more than 110 organisations across over 50 countries.

For Nvidia, the arrangement sits alongside its wider European AI activity. AI Affairs previously reported that the company planned $2 billion in European AI deals after deploying $8 billion. The EBU initiative concerns shared technical requirements for public broadcasters rather than a reported purchase of that size.

Microsoft and Nebius offer different computing routes

Microsoft will contribute expertise from Microsoft Sovereign Cloud, including Azure Local and Foundry Local. Nvidia says those products would allow broadcasters to process sensitive editorial material locally and maintain critical services if external connectivity goes down.

Nebius offers a separate route on European infrastructure. Its platform would let broadcasters train, fine-tune and run models of their own, according to Nvidia. The planned framework encompasses both the location of that computing and the rules governing the data used by the resulting services.

Antonio Arcidiacono, the EBU’s chief technology and innovation officer, said the standards would give members a common basis for retaining control and providing accountable services. The initiative covers potential uses of agents in newsrooms, production and audience services, rather than a single application for all members.

Nemotron and Qvest address models and integration

Nvidia’s Nemotron open models are the proposed starting point for applications tailored to broadcasting. Nvidia says broadcasters can adapt the models to particular domains and public-service requirements. The approach also gives members scope to inspect and evaluate models against local languages, archives and editorial policies.

Delphine Ernotte, chief executive of France Télévisions and president of the EBU, said sovereign cloud infrastructure and open models could support innovation while protecting data and editorial independence. The model component places Nvidia’s software alongside the computing and cloud products in the expanded framework.

Qvest and other systems integrators will provide media-specific consulting and integration. Their role is to make the reference architectures practical for broadcast environments and interoperable with the systems used there. The EBU and Nvidia will define those architectures as part of the expanded initiative announced in Berlin on 19 October.

Sources

Topics: Enterprise adoption, Public sector, Sovereign AI