NVIDIA, SoftBank Collaborate on Japan's Most Powerful AI Supercomputer and AI-5G Integration
In This Article:
NVIDIA (NVDA, Financials) and SoftBank Corp. (SFTBY, Financials) unveiled plans to bolster Japan's AI capabilities by creating the country's most powerful AI supercomputer, which will integrate AI and 5G networks for telecom providers, according to a Tuesday announcement.
Declared during the NVIDIA AI Summit Japan, the partnership highlights NVIDIA's Blackwell platform, meant to support Japan's AI and 5G projects and provide fresh telecom industry revenue sources.
Using the NVIDIA Blackwell platform first then switching to the Grace Blackwell platform, NVIDIA CEO Jensen Huang said that SoftBank is creating its new supercomputer. SoftBank has tested the first AI and 5G integrated network using NVIDIA AI Aerial, hence enabling AI-driven telecom breakthroughs. For every $1 invested on capital expenditure, this AI-5G technology presents telecom businesses with a possible income of $5 in AI inference.
To assist the growth and implementation of AI solutions in Japan, the firms also detailed strategies to create an artificial intelligence marketplace run under NVIDIA AI Enterprise software. SoftBank wants to be an artificial intelligence hub for many different sectors, enabling AI-driven developments in robotics, healthcare, and telecommunications among other ones.
To build the DGX SuperPOD, which is expected to be Japan's top-performing supercomputer, SoftBank's future AI architecture calls for NVIDIA DGX B200 systems. Designed to serve local companies, colleges, and research labs, this infrastructure will also serve generative AI and massive language model initiatives.
Furthermore, SoftBank's network can perform AI and telecom workloads concurrently thanks to NVIDIA's AI-RAN technology, hence transforming telecom infrastructure. Dubbed AI-RAN, an AI-enabled network architecture was effectively tested in Kanagawa, Japan, with SoftBank conducting concurrent AI inference jobs while displaying carrier-grade 5G speed.
This article first appeared on GuruFocus.