- Led by Nvidia, SoftBank, T-Mobile USA, Ericsson, Nokia and others, the AI-RAN Alliance launched earlier this year
- A couple weeks ago, the alliance announced that industry veteran Alex Jinsung Choi was appointed as chairman
- Choi brings over 30 years of experience to the role and most recently served as chairman of the O-RAN Alliance
One of the running jokes at Mobile World Congress (MWC) in Barcelona earlier this year revolved around the use of “AI” on every booth and sign imaginable. It was as if marketers were sticking “AI” on everything to make their messaging stand out, only to be buried with every other AI wannabe.
So it’s not surprising that a group calling itself the AI-RAN Alliance, which made its debut at MWC, was met with some degree of skepticism. But the alliance’s newly installed chairman is here to tell us: It’s serious business and no laughing matter.
“It is crucial to understand how AI technology can effectively address the challenges faced by the telecom industry,” Alex Jinsung Choi told Fierce.
Choi is the former chair of the O-RAN Alliance and he’s now at the helm of the AI-RAN Alliance. He’s also principal fellow of SoftBank Corp.’s Research Institute of Advanced Technology, but you might also know him from his days as head of T-Labs at Deutsche Telekom or as the first chairman of the Telecom Infra Project (TIP). He also was CTO at SK Telecom.
Fierce caught up with Choi to get an update on the AI-RAN Alliance’s progress. He explained that unlike the O-RAN Alliance, the AI-RAN Alliance is not a standards development organization (SDO). However, he will be able to apply his experience from the O-RAN Alliance to lead another diverse set of stakeholders in the AI initiative.
Getting to work
So, what can AI in the RAN do for operators that they can’t do with AI outside of the RAN?
Choi said AI in the RAN offers unique advantages that external AI applications can’t match. By integrating AI directly within the RAN, operators can achieve real-time network optimization – and this is key, he said. External AI solutions may not be able to interact as directly or quickly as AI in the RAN.
He likened it to an automobile company that’s trying to develop a self-driving car – the goal is to make as many driving-related decisions as possible within the car itself rather than relying on a distant cloud. In other words, if an operator can detect a problem within the network faster with AI, it’s going to be far better off than using external solutions.
As for whether or not you need a standalone (SA) 5G network to do AI in the RAN, Choi said the short answer is “yes and no.” Operators can introduce AI applications without having 5G SA. It might not produce the best experiences, but it’s acceptable.
“As long as you have enough bandwidth and computing power,” you can do AI with non-standalone (NSA) 5G," Choi said.
To kick things off, the AI-RAN Alliance created three working groups, and it has now selected leaders to oversee these groups:
AI-for-RAN Working Group: The aim is to explore and enhance the use of AI in RANs to improve performance metrics, such as efficiency and capacity, as the industry moves toward 6G.
AI-and-RAN Working Group: The mission here is to explore the concurrent use of converged computer-and-communications infrastructure to run RAN, AI and Generative AI (GenAI) workloads, enhancing platform utilization and creating new monetization opportunities.
AI-on-RAN Working Group: This group is tasked with defining the radio interface requirements for running AI and GenAI applications across consumer, enterprise and government sectors. The aim is to benchmark the performance of these applications on 5G and identify new requirements for future 6G systems.
AI from the analyst perspective
The AI RAN Alliance has a vision to add intelligence that can bring new value to the RAN – and it has an impressive list of industry players with the expertise and years of experience to do it, said Sue Rudd, managing director at BBand Communications. More on who's involved in a minute.
Rudd noted RAN use of machine learning/AI for self-organizing networks (SON) is probably the most advanced and well proven application of AI in real time telecom operations.
“The AI RAN Alliance is bringing together a critical mass of leading industry players to create an ecosystem that can simultaneously optimize RAN performance AND enable AI applications for edge use cases,” she told Fierce via email.
“AI at the edge” on this type of AI RAN Alliance platform “may be the anchor application that finally triggers large scale deployment of edge services in 2025,” Rudd said.
Who's who
Choi noted it’s crucial for the alliance to understand and communicate how AI technology can effectively address the challenges in the telecom industry while differentiating itself from organizations like 3GPP and the O-RAN Alliance.
The founding members of the AI-RAN Alliance include Arm, DeepSig, Ericsson, Microsoft, Nokia, Northeastern University, Nvidia, Samsung Electronics, SoftBank, the University of Tokyo and T-Mobile USA.
Since the launch, the following members have joined: Aarna Networks, ETRI, Hewlett Packard Enterprise, Institute for Information Industry, Itochu Techno Solutions, ITRI, Keysight Technologies, MatSing, Rohde Schwarz, Seoul National University, Singapore University of Technology and Design, Tietoevry, the University of Oulu, Viavi Solutions and Yonsei University.
In the early days of its launch, some companies openly wondered why they weren’t invited to be part of the AI-RAN Alliance.
Mavenir, for one, was miffed because Nvidia is an investor in Mavenir and didn’t think to extend an invite. This week, a Mavenir spokesperson said the paperwork has been submitted to join the AI-RAN Alliance, so Mavenir is waiting for that to go through before making it official.
Choi said the expectation is that more members will join and that membership is “wide open.”