Nvidia and SoftBank reached a new milestone on November 12 at the Nvidia AI Summit Japan, where CEO Jensen Huang announced the successful testing of an integrated AI and 5G network system.
The announcement follows their earlier collaboration through the AI-RAN Alliance, which they helped establish at the Mobile World Congress in Barcelona last February.
This integrated network dubbed 5G AI-RAN aims to convert traditional cellular base stations from pure infrastructure costs into revenue-generating AI computing hubs.
SoftBank demonstrated this in a real-world trial in Kanagawa Japan, claiming that base stations equipped with Nvidia’s technology could maintain full 5G performance while also running AI workloads using spare network capacity.
According to SoftBank, 5G AI-RAN could harness untapped 5G network capacity for AI processing, with projections suggesting operators could earn $5 for every $1 invested and see returns up to 219% per server.
SoftBank First to Secure Nvidia’s Next-Gen Blackwell Chips
At the summit, Huang revealed that SoftBank is the first to deploy Nvidia’s Grace Blackwell chips. These chips will be incorporated into what SoftBank expects to be Japan’s most powerful AI supercomputer, designed for various AI applications, including generative AI development.
Using Nvidia’s AI Enterprise software, SoftBank said it plans to launch an AI marketplace that will address Japan’s increasing need for localized and secure AI computing. However, no timeframe was provided for this.
This marketplace will support AI training and edge inference, with potential applications across healthcare, transportation, and manufacturing.
The Nvidia-SoftBank partnership marks another step in Nvidia’s growing Asian expansion. In August 2024, the chip maker established an AI research and development center in Taiwan. The company is reportedly readying another huge investment announcement in Thailand next month.