In a groundbreaking move, Ericsson is advancing the monetization of AI in Radio Access Networks (RAN) with an innovative approach that does not rely on GPU hardware. The software subscription model from Ericsson serves as a counter to its competitor Nokia, which has been integrating AI capabilities into its networks using GPUs in partnership with NVIDIA. This strategy by Ericsson potentially offers a cost-effective way for service providers to enhance performance without investing in additional hardware.
Ericsson’s AI in RAN has successfully undergone numerous commercial network deployments worldwide, showcasing notable improvements. The technology reportedly increases downlink throughput by up to 20% and improves spectral efficiency by 10%. Furthermore, it purportedly supports twice the number of high-volume users and offers superior coverage predictions and user positioning accuracy. Joe Madden from Mobile Experts highlights the impressive return on investment, emphasizing how operators can leverage existing 5G networks for improved capacity and services.
The competitive dynamics extend beyond technology. Nokia’s decision to pivot towards an AI infrastructure model following its partnership with NVIDIA has reflected positively in its stock performance. Despite this, Ericsson is confident that its software-centric strategy will also capture market interest.
In an attempt to highlight its continued commitment to innovation, Ericsson has been embedding AI in its products since the 5G era. Earlier this year, it launched AI-ready radios that significantly enhanced inference capabilities. This foresight manifested in its current AI in RAN, which integrates telco-grade AI models, designed to run in real-time, into basebands and radios. This offering is commercially scalable and promises immediate benefits for 5G networks, allowing for operational and energy efficiencies without hardware upgrades.
Operators are increasingly taking note of these advancements. Teruyuki Oya from SoftBank Corp emphasizes how these developments allow for real-time radio performance optimization, enhancing spectrum efficiency and user experience. Bruce Dean from Bell appreciates the move towards more intelligent and efficient networks, while Yu Takki from SK Telecom points to research and software innovation as key to future AI network evolution.
Through collaborative initiatives with industry giants, Ericsson is setting the stage for AI-native networks, promising enhanced performance and efficiency. Mårten Lerner from Ericsson encapsulates this vision, indicating the transformative potential of AI to redefine mobile networks’ capabilities and operational efficiencies.
As the telecom industry evolves, Ericsson’s approach offers an intriguing alternative, potentially balancing high performance with economic viability. This could catalyze a shift towards more widespread AI integration across networks, marking a significant step towards the future of intelligent telecommunications infrastructure.
