SANTA CLARA — In a groundbreaking collaboration, NVIDIA and SoftBank Corp. have unveiled plans to deploy a pioneering platform tailored for generative AI and 5G/6G applications. This innovative endeavor hinges on the integration of the NVIDIA GH200 Grace Hopper™ Superchip within SoftBank’s next-generation data centers, slated for deployment across Japan. Leveraging NVIDIA’s advanced computing architecture, SoftBank aims to establish a distributed network of AI-driven data centers capable of hosting a myriad of generative AI and wireless applications. These data centers, designed in partnership with NVIDIA, promise enhanced performance, scalability, and resource utilization, marking a significant milestone in the evolution of data center architecture.

Junichi Miyakawa, President and CEO of SoftBank Corp., emphasized the critical role of next-generation infrastructure in supporting Japan’s super-digitalized society. Miyakawa underscored the transformative potential of AI-driven optimization, particularly in Radio Access Network (RAN) operations, aiming to reduce energy consumption and facilitate resource sharing across interconnected data centers.

Jensen Huang, Founder and CEO of NVIDIA, hailed the NVIDIA Grace Hopper platform as a revolutionary computing paradigm poised to reshape the landscape of generative AI services. Huang lauded SoftBank’s visionary leadership in pioneering telecom networks primed for hosting AI-driven services, reaffirming NVIDIA’s commitment to driving innovation in accelerated computing.

The deployment of these new data centers signifies a strategic shift towards a more distributed infrastructure, poised to handle a diverse array of AI and 5G workloads with enhanced efficiency and reduced latency. SoftBank’s ambitious agenda includes the exploration of 5G applications spanning autonomous driving, AI factories, augmented reality, computer vision, and digital twins.

In a bid to further accelerate 5G innovation, NVIDIA’s Grace Hopper and BlueField®-3 data processing units will power the software-defined 5G virtual RAN, alongside generative AI applications. This collaboration seeks to deliver unprecedented throughput and precision timing protocols for 5G networks, revolutionizing the capabilities of industry-standard servers.

Rene Haas, CEO of Arm, highlighted the pivotal role of energy-efficient compute solutions in driving the future of generative AI. Haas praised the synergy between NVIDIA’s Grace Hopper Superchip and BlueField DPUs, citing their potential to unlock exponential efficiency gains for software-defined 5G and AI applications on Arm architecture.

With the convergence of NVIDIA’s advanced computing technologies and SoftBank’s visionary approach, the stage is set for a new era of innovation in data center architecture, fueling the advancement of generative AI and wireless communication technologies.