Schneider Electric has unveiled its new Galaxy VXL, an uninterruptible power supply (UPS) system designed to meet the growing power demands of artificial intelligence and data centre operations.
The compact 3-phase UPS delivers up to 1.25 MW of power in a single frame and can scale up to 5 MW with four units operating in parallel.
Galaxy VXL occupies just 1.2 square metres of floor space, representing a 52 per cent smaller footprint compared to industry average systems, while achieving power density of up to 1042 kW per square metre.
Furthermore, the Galaxy VXL is compatible with both lithium-ion and VRLA (Valve-Regulated Lead-Acid) batteries.
The system operates with up to 99 per cent efficiency in its eConversion mode and up to 97.5 per cent in double conversion mode, which the company claims reduces carbon emissions by half compared to conventional systems.
Designed with modularity in mind, the UPS uses 125 kW power modules that can be added incrementally as power requirements grow, allowing customers to implement a pay-as-you-grow strategy.
For enhanced reliability, the system features redundancy capabilities that increase availability by a factor of 10, according to Schneider.
It also includes Live Swap technology enabling power module replacements without system downtime.
"As dependency on large-scale infrastructure systems grows at an unprecedented rate, it's crucial that customers can access the most sustainable, resilient, and efficient technologies to safeguard their critical systems, while minimising their energy and environmental impact," Joe Craparotta, vice president of cloud and service providers at Schneider Electric, said.
Digital twin for AI power requirements launched
Earlier on, Schneider and its partners launched a digital twin that simulates artificial intelligence factory power requirements.
The company worked together with United States power system design company ETAP, using the Nvidia Omniverse platform.
Taking a "grid to chip" approach, the digital twin addresses the challenges of power management, performance optimisation, and energy efficiency for AI facilities.
Data centre operators can currently estimate average power consumption at the rack level, but with the new digital twin, increased precision on modelling dynamic load behaviour at the chip level can be done.
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
