Open RAN is entering a more demanding phase as artificial intelligence expands across mobile networks. The technology promises wider vendor choice and more flexible network design. Yet it also places new pressure on interoperability, security, and testing.
That message emerged during the RCR Wireless News Intelligent RAN Forum 2026. Viavi Solutions marketing manager Owen O’Donnell and Battelle vice president Mark Reudink discussed the changing radio access network landscape.
O’Donnell described openness and intelligence as a powerful mix for operators. “I think, in a way, openness and network intelligence is nearly a perfect storm for the operators because the openness is bringing in new players, it’s increasing the playing field, it’s reducing the cost, while network intelligence is allowing operators to head towards the goal of fully autonomous networks,” O’Donnell said.
This shift could lower capital and operating costs. It may also speed service launches and simplify daily operations. For operators, that means more room to innovate. For vendors, it opens the door to new products and specialized applications.
However, the model also adds complexity. Operators must connect hardware and software from multiple suppliers. Each component must work reliably with the others. That challenge grows as AI applications start making real network decisions.
Reudink said integration has improved sharply. Battelle’s first active antenna radio unit integration took months. Similar work can now take weeks. He credited better vendor software and more mature testing platforms.
At the same time, new use cases continue to appear. These include network steering and indoor location services. Such tools can help factories where GPS signals do not work well.
The RAN Intelligent Controller, or RIC, has also moved forward. The RIC helps manage applications that optimize radio network behavior. O’Donnell said it has moved beyond paper specifications into production lab testing.
Current use cases include network-slice assurance and massive MIMO optimization. Others include signaling-storm detection, drone-swarm detection, and application-collision testing. These tasks require dependable data and careful validation.
O’Donnell also cited a demonstration with NTT Docomo. It used AI-driven beamforming control to reduce control overhead. The approach could improve system throughput by up to 20%.
Still, O’Donnell warned that testing remains central. “The challenges are interoperability, integration, compliance and standards, security, increased attack surface, and regular software updates,” he said. “So they’re all the primary challenges that come with Open RAN.”
Trust in AI remains another major issue. AI models depend heavily on training data quality. Yet real operator data can be sensitive, private, and difficult to share.
Synthetic data may offer one answer. Viavi’s AI RAN Scenario Generator can create network scenarios. It can also train AI models and validate their performance.
Reudink stressed that vendors need common data approaches. “The data that comes in is fundamental to be able to have accurate AI models, to be able to really implement changes in that network, be it anything from smarter beam management to network steering,” he said.
Looking ahead, the AI-RAN Alliance could help align vendors and operators. Standard metrics and shared testing methods will matter more. Multi-vendor labs will help operators compare competing AI tools before deployment.
