Nokia and Nvidia expand sovereign AI pact for telecom networks

Nokia and Nvidia have expanded their partnership to build a sovereign agentic AI platform that telecom operators can use to automate network operations while keeping control of their own data and infrastructure. Three operators — etisalat UAE (formerly e&), MEO and Airtel — are already working with Nokia to explore potential applications for the technology.

That operator interest matters more than the press release suggests. Vendors can announce platforms; carriers signing on to test them is what turns marketing into a roadmap.

What will the platform actually do?

The platform combines Nvidia AI Enterprise software, Nemotron models and libraries with Nokia's network technology. Operators would use it to build and manage AI agents across four domains: radio, core, transport and cloud networks.

According to Nokia, these agents will monitor network conditions, identify problems and take action to improve performance, reducing the need for manual intervention. In other words, the network starts fixing itself before an engineer opens a ticket.

Nokia says the system will include safeguards that let operators oversee how AI agents make decisions and carry out tasks. Those guardrails are the sovereign part of the pitch — oversight, not autonomy by default.

Deployment flexibility is part of the design. The platform will support on-premises, cloud and hybrid deployments, giving operators control over where their operational data, models and policies are stored. For carriers with data-residency obligations or nervous regulators, that's not a footnote. It's the product.

What did the executives say?

Nokia Chief Technology and AI Officer Pallavi Mahajan framed the ambition in expansive terms. "The next era of connectivity will be defined by networks that do far more than connect: They will distribute intelligence, continuously optimize performance and enable entirely new AI-powered services, including the emerging world of physical AI," she said.

She added: "Together with Nvidia and leading operators, Nokia is turning this vision into reality by bringing AI into the radio network and across network operations. This creates a trusted, software-defined foundation that helps operators get more from their spectrum today while enabling new Physical AI use cases and opening a practical path toward increasingly autonomous, AI-native 6G networks."

Nvidia's Senior Vice President of Telecom, Ronnie Vasishta, kept his claim tighter to the architecture. "The next generation of telecommunications infrastructure will be built on a software-defined architecture," he said.

"Nvidia and Nokia are combining CUDA accelerated computing and AI with Nokia's network software and expertise to improve radio performance and deploy trusted AI agents. Together, we are advancing software-defined, AI-native 5G and laying the foundation for 6G," Vasishta added.

Note the verbs. Both executives describe advancing and laying foundations — not shipping. No commercial availability date, pricing or deployment figures appear in the announcement.

How does this fit the AI-RAN work?

The announcement builds on the companies' wider AI-RAN partnership, which is already being tested by operators including Elisa, Deutsche Telekom, BT and Orange. That earlier work put AI compute into the radio access network; this expansion extends the idea from the radio layer to agents managing operations across the whole network stack.

The distinction between the two operator groups is worth watching. Elisa, Deutsche Telekom, BT and Orange are testing AI-RAN. Etisalat UAE, MEO and Airtel are exploring the agentic platform. Whether those tracks converge into single deployments, or remain parallel vendor experiments, will determine whether this becomes an operating model or a conference slide.

Sovereignty is the commercial hook. European and Gulf operators face growing pressure to keep subscriber data and network intelligence within national borders, and a platform that runs on-premises with auditable agent behaviour answers that pressure more directly than a public-cloud alternative.

The 6G framing sets the timeline. Both companies position the agentic platform as a stepping stone toward AI-native 6G networks, which means the real test arrives with the first commercial 6G standardisation cycle later this decade — with the nearer-term question being whether etisalat UAE, MEO and Airtel move from exploring to deploying.