SINGAPORE: Singtel is launching GPU-as-a-Service (GPUaaS) in Singapore and South-East Asia in the third quarter of 2024.
GPUaaS is a cloud-based solution that offers access to graphics processing units (GPUs) on demand.
GPUs are essentially a type of artificial intelligence (AI) chip that is most often used in the training of AI models. Their potential applications include machine learning, graphics rendering, data processing, high-performance computing and gaming.
On March 19, Singtel said its GPUaaS will be deployed through Nvidia H100 GPU-powered clusters that are operated in Singtel’s existing upgraded data centres in Singapore.
The telecommunications provider intends to eventually expand its GPUaaS to run in its three upcoming sustainable AI data centres across Singapore, Thailand and Indonesia when they begin operations.
These are currently being developed by the group’s regional data centre business, Nxera. Singtel plans to deploy over 100 megawatts of such data centres through Nxera within the next three years.
The company will be among the world’s first to deploy Nvidia’s GB200 Grace Blackwell Superchips upon receiving them in early 2025.
Singtel said this would give its enterprise customers options for different types of accelerators for their advanced computing and AI needs.
The Blackwell GPUs are said to deliver 30 times faster real-time large language model inference than their predecessors, as they use over 200 billion transistors.
Bill Chang, chief executive of Singtel’s Digital InfraCo unit as well as Nxera, said: “Our GPUaaS will run in AI-ready data centres specifically tailored for intense compute environments with purpose-built liquid-cooling technologies for maximum efficiency and lowest power usage effectiveness, giving (enterprise customers) the flexibility to deploy AI without having to invest and manage expensive data centre infrastructure.”
He added that the group has been seeing “keen interest” from both private and public sectors, which are “raring to deploy AI at scale quickly and cost-effectively”.
Separately, on March 19, Singtel announced that it launched a digital aggregation and orchestration platform, Paragon-S, at the global Satellite 2024 Conference & Exhibition in Washington.
Paragon-S was developed in-house and is based on Singtel’s patented orchestration platform Paragon. It integrates multiple orbit satellites with terrestrial fixed and mobile network connectivity services.
It enables enterprises to switch applications between satellite operators’ edge cloud and public clouds based on their business requirements.
Singtel said this marks the satellite industry’s first all-in-one orchestration platform, which it believes will help satellite operators enhance resiliency and optimise cost, while also unlocking new streams of revenue through edge computing and AI.
“The introduction of Paragon-S will empower satellite operators to become platform companies for enterprises. We will build on our existing ecosystem with partners to help operators develop and accelerate time to market for their use cases,” commented Chang. — The Straits Times/ANN