Japanese software service provider Datasection has signed an agreement with Inventec to purchase 1,250 servers equipped with 10,000 Nvidia B300 GPUs. The company plans to build a world-class, first-of-its-kind large-scale AI cluster featuring Nvidia B300 GPUs in Sydney, Australia.
Datasection president and CEO Norihiko Ishihara stated that the cluster's computing power will exceed 200 ExaFLOPS (FP4) to meet the demands of one of the world's largest cloud service providers.
This supercluster uses Nvidia's latest Blackwell B300 GPU to deliver enterprise-grade AI computing services for businesses across Japan and the Asia-Pacific region. The initiative aims to address the severe shortage of high-performance AI computing infrastructure as market demand far outstrips supply, with enterprises and public institutions accelerating their AI transformation projects.
Strategic partnership accelerates deployment
Ishihara highlighted that through strategic collaboration with Inventec, Datasection secured advanced and highly reliable GPU servers to deploy and scale its AI data center operations. This partnership enables Datasection to gain priority access to next-generation hardware, accelerating expansion across multiple markets.
World's first ultra-large B300 implementation
This project marks a significant step for Datasection in solidifying its position as a leading and trusted AI cloud service provider in the Asia-Pacific region. The Sydney-based AI cluster is a massive deployment comprising 10,000 GPUs across 1,250 units—the world's first ultra-large-scale implementation using Nvidia B300 GPUs.
Ishihara stated that the company is excited to share this achievement globally and looks forward to continuing close cooperation with partners to rapidly build infrastructure that powers next-generation AI innovation in the Asia-Pacific. In addition to Inventec, Datasection recently announced the procurement of 625 servers equipped with 5,000 Nvidia B200 GPUs from Giga Computing.
Proprietary TAIZA algorithm optimizes performance
Datasection's core competitive advantage lies in its proprietary algorithm system, TAIZA, which optimizes the operation of large GPU clusters under AI workloads. Compared to standard cluster management solutions, TAIZA significantly improves resource utilization, reduces operational costs, and enhances performance reliability.
Designed to maximize efficiency in AI data center architectures, this technology delivers optimal performance for machine learning training, inference workloads, and high-performance computing applications—offering customers a clear and compelling value proposition.
Capturing Asia-Pacific market growth
Ishihara noted that, driven by accelerated AI adoption in manufacturing, financial services, healthcare, and technology sectors, the Asia-Pacific AI infrastructure market is expected to grow substantially. Datasection aims to leverage its early-mover advantage in B200/B300 technologies to capture a significant market share in this expanding sector.
Article edited by Jerry Chen
