YTL Power International has established YTL AI Cloud, a specialised provider of massively scaled graphics processing unit (GPU)-based accelerated computing. Its primary objective is to deploy and oversee one of the most advanced supercomputers globally, powered by Nvidia’s Grace Blackwell architecture.
This artificial intelligence (AI) supercomputer, known as Nvidia Grace Blackwell-powered DGX Cloud, aims to accelerate the development of generative AI, according to The Edge Malaysia.
YTL Power stated that it stands among the pioneering adopters of the Nvidia GB200 NVL72, a liquid-cooled, rack-scale system comprising multiple nodes. These nodes are interconnected through the Nvidia Quantum InfiniBand networking platform.
The platform functions akin to a singular GPU, boasting 1.4 exaflops of AI performance and 30TB of high-speed memory, tailored for the most computationally intensive tasks. The YTL AI Supercomputer is poised to exceed 300 exaflops of AI compute, positioning it as one of the fastest supercomputers globally.
Nvidia’s founder and CEO, Jensen Huang, highlighted the collaboration with YTL AI Cloud, aiming to introduce a top-tier accelerated computing platform to Southeast Asia. This initiative is anticipated to propel scientific research, foster innovation, and drive economic expansion throughout the region.
The YTL AI Supercomputer will be housed within the expansive YTL Green Data Centre Campus in Johor, spanning 1,640 acres (663.68 hectares) and powered by renewable energy sourced from its on-site 500MW solar facility.
This strategic placement aims to cater to the rising demand for highly scalable, high-performance cloud-based solutions tailored for AI and machine learning tasks.
