Softbank telco AI strategy includes working with NVIDIA on network automation and adding AI overhead into 5G RAN compute
As interest and investment in artificial intelligence has reached a fever pitch, NVIDIA has emerged as the undisputed leader in using its accelerated compute architecture and portfolio to bring the power of AI into a wide range of verticals, including via is Telecom business unit. In conversation with RCR Wireless News, NVIDIA SVP of Telecom Ronnie Vasishta gives an overview of the applicability of accelerated compute to operators, as well as the internal- and customer-facing use cases supported by telco AI capabilities.
In a survey released in February, NVIDIA found that nearly all telecommunications respondents (95%) are engaged with AI. However, much of this engagement is nascent, as only 34% of respondents reported using AI for more than six months, and 23% said they’re still learning about the different options for telco AI applications. A further 18% are in the trial or pilot phase. According to the survey, 65% of respondents agreed that AI is important to their company’s success, with 60% expecting AI to help them optimize operations, 44% expect AI to help them reduce costs, 35% believe AI will enhance customer engagement and 31% think it will help them meet revenue targets.
Earlier this year NVIDIA and Softbank announced a telco AI collaboration that would use the former’s GH200 Grace Hopper Superchip in the latter’s distributed data center footprint in Japan. The idea is to build telco AI data centers that apply AI to the operation of the 5G radio access network while also creating a multi-tenant server platform to host genAI applications.
Also in Japan, NTT Docomo is deploying a GPU-accelerated telco AI solution in its network in Japan, reportedly making it the first operator in the world to adopt a GPU-accelerated commercial 5G network and representing a notable milestone in delivering Open RAN’s promise of flexibility, scalability and innovation.
Further, the network is also the first 5G vRAN for telecom commercial deployment to use the Nvidia Aerial platform, which combines a virtualized radio access network (vRAN) stack for 5G, telco AI frameworks, accelerated compute infrastructure and long-term software support and maintenance. On top of the Nvidia Aerial platform, sits a virtualized centralized unit (vCU) and a virtualized distributed unit (vDU) from Fujitsu and Wind River’s distributed cloud platform.
