The RAN is no longer just a connectivity layer – it is becoming an intelligent, programmable platform. At MWC26, a session moderated by Abe Nejad of The Network Media Group (NMG) brought together leaders from AT&T, RadiSys, Andrew, Aira Technology, and Rakuten Symphony to examine what it actually takes to get there. The conversation covered the full arc: from the maturity of operator-supplier collaboration and the role of Open RAN, to new revenue models unlocked by AI and the hard realities of scaling autonomous operations.
Speakers:
Watch the full interview.
The panel opened with a candid assessment of where operator-supplier collaboration stands today. The conversations are more substantive than they were a year ago: standards bodies like TM Forum are providing the structural scaffolding operators need to align on automation requirements, but a critical gap persists. Operators are sitting on vast stores of network data, yet only a fraction of it is being put to work. The challenge is not just collecting data; it is organizing it, governing access to it, and making it available in a form that allows AI models to be meaningfully trained and tested. The leaders were direct: until that data infrastructure matures, the most ambitious automation goals will remain out of reach.
On the Open RAN front, the panel moved past theory and into deployment specifics. Operators like AT&T have established commercial traffic on open small cell radios, brought Cloud RAN live in the network, and are actively using rApps to provision and manage everything from third-party radios to DAS infrastructure. The panelists emphasized that the application layer is opening up to genuine third-party innovation for the first time. For indoor deployments specifically, Open RAN is enabling full end-to-end network visibility that simply did not exist under legacy architectures, and multi-vendor interoperability, while still a learning process, is getting meaningfully easier with each successive deployment.
One of the more forward-looking threads in the discussion was around monetization. The panel agreed that AI's near-term value in the RAN is largely operational – energy savings, outage compensation, predictive maintenance – but the longer-term opportunity is commercial. When operators can tap into a larger share of their network data at scale, they gain predictive insight that goes beyond efficiency: understanding where demand will spike, how users will move, and what experience a device will encounter in a given location. New API exposure models, built on AI-derived network intelligence, represent a credible path to revenue generation.
The panel's most grounded moments came when the conversation turned to industrialization. Finding AI use cases is not the problem; the industry has no shortage of them. Running those use cases reliably across a nationwide network is an entirely different challenge. The leaders stressed that robust data pipelines, tight feedback loops, and the ability to roll back changes quickly are not implementation details; they are the foundation on which any serious AI deployment stands. The consensus was direct: AI in the RAN is an evolution, not a revolution, and the operators who will lead are those who invest in making their stack resilient enough to support it at scale.
“Finding the use case is the easier part. The harder thing to do is to industrialize it at a scale where it can run across a country. We have to make sure that the data that we’re getting is good. And if something goes wrong, you want to be able to turn it off or roll back... AI is not a revolution. It's an evolution."
