Ericsson and SoftBank are advancing their collaboration in Artificial Intelligence (AI) within the Radio Access Network (RAN) sector. This partnership has already introduced a prototype of Ericsson’s Cloud RAN software leveraging NVIDIA‘s cutting-edge GH200 Grace Hopper Superchip architecture. The duo has also defined a novel architecture that co-orchestrates RAN and AI applications on shared infrastructure. Their achievements signal substantial progress in integrating AI with RAN, promising enhanced networks.
The two leaders plan to investigate offloading RAN functions to the Graphics Processing Unit (GPU), using NVIDIA’s Grace Hopper Superchip. This technology could optimize performance, allowing for more efficient network operation. Moreover, they are working on orchestrating AI and RAN workloads by combining SoftBank’s AITRAS orchestrator and Ericsson’s open network management platform. This endeavor aims to optimize AI handling in robotics by leveraging Ericsson’s compute offloading framework.
Jawad Manssour, the President at Ericsson Japan, emphasized their dedication, stating: “The progress made in our collaboration with SoftBank underscores our commitment to pioneering innovations in telecommunications.” Their focused efforts reflect a strong commitment to advancing AI-RAN integration.
On the other hand, Hideyuki Tsukuda, Executive VP at SoftBank, remarked on the importance of building SoftBank’s infrastructure with AI-RAN. This strategy is integral to developing next-generation communication infrastructure. Their joint research aims to create networks that efficiently support evolving AI technologies.
The duo extends their vision through the AI-RAN Alliance, a collaborative group aiming to merge AI with cellular networks. Debuted at MWC Barcelona 2024, the alliance aims to establish a common path for AI-driven advancements globally.
