Open radio access network (RAN) technology remains a work in process due to complexity and integration challenges, but an increasing number of operators and vendors are getting behind AI as a way to tackle some of that complexity and wring more efficiency from their RAN network deployments.

Analysts have pointed to the benefits AI can bring to the deployment and management of cloud-based open RAN networks, which are more complex orchestration challenges due to the disaggregated multivendor ecosystem. This includes the use of AI to help close performance gaps for open RAN architectures compared with legacy RAN models.

“The integration of AI and ML (machine learning) techniques, along with other innovations in energy efficiency and GPU acceleration, will accelerate performance improvements closer to traditional RAN networks,” ABI Research open RAN research analyst Larbi Belkhit noted in a report. “This will remove critical barriers to open RAN adoption and pave the way for flexible, interoperable 5G deployments for network operators rather than reliance on radio network equipment from traditional vendors currently dominating the market, such as Ericsson, Huawei, and Nokia.”

Subhankar Pal, senior director for innovation at Capgemini Engineering, in an interview with SDxCentral explained that one highly touted use case is the capability to use AI to better manage network resources. This involves being able to actively manage the use and power requirement of a cell site.

“Switching of a cell or switching off some of the antennas at night when they’re required, or even when we are speaking, maybe there are pauses, intermediate pauses, and can some of the time slots or the communication be shut off at the time,” Pal said.

This might not seem like a significant benefit, but operators typically have tens of thousands of large cell sites on their network, with ongoing deployment of smaller cell sites designed for more targeted and localized coverage needs. Having these all running when not needed is a huge drain on energy and financial resources.

As an example, Pal said larger carriers can spend up to $1.5 billion per year just on energy.

“If you can reduce 10% of this energy, that’s a huge cost saving and of course also helping your net-zero targets,” Pal said. “Energy saving is a big area for this.”

AI to control the RAN

Operators have latched onto this opportunity, with the most prominent orbiting around the AI-RAN Alliance.

That organization was launched in early 2024, when a handful of big-name vendors and operators threw their support behind the aptly named AI-RAN Alliance organization. Those names included Amazon Web Services (AWS), Arm, Ericsson, Microsoft, Nokia, Samsung Electronics, SoftBank, Nvidia, DeepSig, T-Mobile US, and Northeastern University.

The framing goal of the group is to steer the use of AI into RANs for better performance, lower operating costs, greater efficiencies, and to support new business models. That work was to include using AI to improve RAN spectral efficiency, combine the two for more efficient network utilization, and deploy AI at the network edge to support new services.

Those goals did progress over the proceeding 12 months, with a number of operators and vendors pushing AI-RAN agendas.

One of the more notable moves came late last year when T-Mobile US partnered with Nvidia, Ericsson, and Nokia on an AI-RAN Innovation Center that will house a focused effort on tying together cloud-based RAN and AI development. The carrier noted in a presentation that the goal is to integrate cloud-based RAN and AI using unified infrastructure that can scale to serve millions of mobile users at once.

“AI-RAN will enable new AI algorithms to unlock the full potential of wireless networks,” T-Mobile US’ presentation noted. “These AI algorithms would be rapidly developed with software-defined RAN, trained on AI data centers, and fine-tuned with physically accurate digital twins. This will lead to dramatic improvements in spectral and energy efficiencies.”

ABI Research noted in a recent report that AI-RAN will generate more than $6.1 billion in investment revenues by 2032, though most of that is not expected to begin until 2029. However, the analyst firm also highlighted the dearth of operators among AI-RAN Alliance members, which “signals some industry skepticism regarding the near-term value of AI-RAN.”

“The real growth in AI-RAN will only come when performance benchmarks are validated in the field,” ABI Research analyst Samuel Bowling wrote. “Operators need evidence that AI-RAN can deliver technical and financial outcomes at scale, with a justifiable cost model. Without that, adoption will continue to lag behind vendor enthusiasm.”

AI running the RAN in action

Despite the discrepancy, some operators are taking the plunge.

AT&T, for instance, recently implemented a network wide generative AI tool to help control network operations, a move that the carrier said is a step toward greater network autonomy.

Raj Savoor, VP of network analytics and automation at AT&T, explained in a blog post that the AT&T Geo Modeler system is active network wide. It uses synthetic data and a network foundation model to simulate and predict network coverage “under dynamically changing network and environmental conditions.”

“Using ray tracing, the Geo Modeler simulates radio transmissions in complex geospatial environments and integrates with multiple internal systems to help automate network decisions and change how the network is configured to improve overall performance and help manage connectivity nationwide, even during extreme events or as the seasons change,” Savoor wrote.

Canadian operator Telus is taking that AI opportunity one step further by using AI to power the RAN intelligence controller (RIC).

The deployment involves Telus using Samsung’s RIC platform and accompanying applications from Samsung’s CognitiV Network Operations Suite (NOS). The initial list of applications includes the KPI Anomaly Detector and RAN Anomaly Insight to proactively identify and analyze network issues; the Energy Saving Manager for dynamic traffic prediction and automated orchestration of energy saving features; and the Load Balancing Manager to optimize resource utilization and improve network performance.

The RIC operates in the middle of an open RAN deployment as a kind of linkage to the wider network. It allows operators to deploy near-real time (xApps) and non-real time (rApps) that then allow operators to design and control RAN functions, providing administrative RAN sovereignty over functions that are typically implemented as proprietary features on base stations.

These “apps” are microservices-based applications operating in xApps and rApps that provide an operator with more control over their open RAN environments.

“Adding AI power and the detection is super important for us because we spend a lot of time trying to find things to improve, things that are not working optimally, and it's like looking for a needle in a haystack,” Bernard Bureau, VP of wireless strategy and 5G services at Telus, explained to SDxCentral.

Bureau noted that AI can help with network configurations to virtually self-heal should that needle-like problem be found.

“If there's a power outage or a fiber cut, for example, on the site, how does the rest of the network need to be to react for these things?” Bureau said of that configuration challenge. “With AI playing an important role, it's going to enable us to be extremely precise in how we configure our network.”

That precision can be the key for operators planning to spend tens-of-billions-of-dollars on open RAN deployments. But analysts note more work will need to be done in order to drive confidence in AI.

“Operators need more than hype, they need transparent, validated evidence that AI-RAN delivers real-world performance and long-term value,” ABI Research’s Bowling added. “The AI-RAN Alliance must now move beyond vision statements and facilitate real pilot deployments with tier-one operators, standardize benchmarks, and publish clear comparisons across GPU, CPU, and custom silicon solutions. Demonstrating cost savings and performance in urban, rural, and remote environments will be critical to building trust and moving from trials to widespread commercial deployments by 2030.”

This article first appeared in the SDxCentral Open RAN Supplement. Register below to read the whole supplement for free