Singtel has partnered with Nvidia to establish an artificial intelligence (AI) centre of excellence (CoE) aimed at helping organizations navigate the challenges of infrastructure and skill shortages that hinder the scaling of AI initiatives. This multimillion-dollar facility, announced today, is designed to facilitate the transition of AI projects from pilot stages to full-scale production.
Bill Chang, CEO of Singtel Digital InfraCo, the telecommunications company’s digital infrastructure division, emphasized that this CoE distinguishes itself from others in Singapore through its focus on applied AI. Unlike traditional centers, Singtel’s CoE encourages enterprises to present real-world challenges and collaborate with a network of large language model (LLM) developers, application providers, and systems integrators.
To further support clients in enhancing their AI implementations, the CoE is structured as a testbed that replicates Singtel’s commercial infrastructure. Chang compared it to a national power grid, with AI data centers acting as generators, fixed networks functioning as transmission lines, and edge locations serving as substations. “Think about this centre of excellence for applied AI as a micro AI grid,” he stated. “Not only do you experiment and solve bottlenecks, but when you go for full-scale deployment, you seamlessly flip over to the main AI grid and get the resources.”
Marc Hamilton, senior vice-president of solutions architecture and engineering at Nvidia, noted that the collaboration builds on the five-layer foundation for AI deployments proposed by Nvidia CEO Jensen Huang. This framework includes physical land, power and data center facilities from Singtel’s Nxera datacentre division, Nvidia’s graphics processing units (GPUs), broader AI infrastructure encompassing networking and cloud orchestration, AI models, and finally, applications.
Hamilton also indicated that Nvidia intends to leverage its network of 40,000 AI startups to assist Singtel’s customers in crafting AI applications. He highlighted the significance of open models, such as Nvidia’s Nemotron, which are crucial for maintaining Singapore’s competitive edge and ensuring that “Singapore’s data and unique competitive advantage stays in Singapore, and controlled by Singapore companies.”
In tandem with these efforts, Nvidia will collaborate with Singtel to prepare data centers for extreme power densities while adhering to stringent sustainability standards. Currently, Singtel’s Nxera data centers utilize Nvidia GB200 Blackwell systems operating at approximately 200 kW per rack. However, Chang revealed that the CoE is already setting the stage for the deployment of Nvidia’s next-generation Rubin Ultra chips, anticipated between 2027 and 2029.
“The next generation I’m talking about is 600 kW to one megawatt per rack,” he explained. “That’s insane in terms of power density – 60 to 100 times more than the average data center today.” Hamilton added that this exceptional density minimizes the required data center footprint. Thanks to the enhanced energy efficiency of Blackwell chips, which are 50 times more effective at executing AI models than the previous Hopper generation, substantial computational power necessitates significantly less physical space. “The GPU data center of today, versus many football fields, is much more like a basketball field,” he remarked.
The efficiency of AI data centers will play a pivotal role as the Singapore government prepares to tighten regulations on the sector. A new Digital Infrastructure Act is expected to be introduced in parliament later this year, aiming to set baseline energy efficiency standards for all data centers, alongside mandatory cybersecurity measures and incident reporting protocols for major cloud service providers and data centers.
Beyond supporting enterprises in deploying and scaling AI applications within data centers, the CoE will also emphasize edge AI and low-latency 5G networks as the industry progresses from generative AI to applications in embodied and physical AI, such as autonomous robotics, humanoids, and drone swarms. The CoE will be situated in the Punggol Digital District, an area currently evolving into a precinct-scale testbed designed to assist companies in trialing and commercializing real-world robotic applications in active operational environments.
Chang mentioned that the CoE is projected to commence operations in approximately three months. In the meantime, large clients across sectors including government, healthcare, banking, and transportation are already exploring use cases and utilizing Singtel’s existing GPU resources to initiate their AI endeavors.
See also
Tesseract Launches Site Manager and PRISM Vision Badge for Job Site Clarity
Affordable Android Smartwatches That Offer Great Value and Features
Russia”s AIDOL Robot Stumbles During Debut in Moscow
AI Technology Revolutionizes Meat Processing at Cargill Slaughterhouse
Seagate Unveils Exos 4U100: 3.2PB AI-Ready Storage with Advanced HAMR Tech
