SoftBank and NVIDIA have expanded their partnership to build AI native telecommunication networks across Japan. It is a massive bet. The initiative combines GB200 hardware and telecom models to turn standard mobile networks into distributed intelligence hubs. But the companies have yet to share a timeline for the commercial rollout.
The hardware integration is already underway inside the Japanese carrier network. SoftBank is currently deploying the full stack of computing hardware from the silicon giant. This includes RTX PRO based AI RAN technology and the AI Aerial platform. They are also deploying Nemotron based large language models trained specifically for telecom workloads. The goal is processing power at the base station. This setup allows the network to handle massive AI tasks locally.
Local software development is also central to the agreement. SoftBank plans to run open models alongside Sarashina, a homegrown generative AI series built by SB Intuitions. The partnership is also pushing into physical robotics. They are working with Yaskawa Electric Corporation to integrate NVIDIA Cosmos and Isaac GR00T software into industrial machinery. The work aims to bring advanced automation to Japanese factory floors.
