Singtel is developing a new generation of artificial intelligence (AI) data centres to meet the growing global demand for AI adoption across various industries.
These AI DCs are designed to handle high power density AI workloads with GPU clusters, which are critical for processing AI but consume a lot of power and generate substantial heat.
The new data centres will also feature efficient cooling solutions, real-time server monitoring platforms, and connectivity to Singtel's fixed and 5G mobile networks through its Paragon platform.
Singtel said this setup will facilitate the connection of smart devices to AI clusters, enabling low-latency AI applications at the edge.
DC Tuas, Singtel's latest data centre under construction with a 58MW capacity will be the first of these fourth-generation facilities.
Bill Chang, chief executive officer of Singtel’s digital InfraCo unit said DC Tuas is designed to mitigate carbon impact through the use of lower carbon embedded materials in construction, while driving greater energy efficiency by incorporating liquid cooling capabilities.
It aims to achieve a Power Usage Effectiveness (PUE) of around 1.23 at full load, making it one of the industry’s most efficient data centres.
Singtel is also developing two other state-of-the-art data centre projects in Thailand and Indonesia, expanding its total capacity beyond 200MW from its current operational capacity of 62MW in Singapore.
The company has secured a S$535 million green loan for its current operational capacity and plans to scale up its regional data centre business with an investment from KKR.
Singtel plans to offer various core services in these AI DCs, including cloud co-location, managed hosting, and GPU-as-a-Service (GPUaaS).
Since 2021, Singtel has deployed GPUaaS at the 5G@Sentosa testbed, enabling enterprises and government agencies to build AI applications without incurring significant capital expenditures.
