Delta has introduced a 5,500-W redundant power supply unit designed for AI data centers. It supports both 19-inch (1RU) and ORv3 21-inch (1OU) rack configurations. This unit recently received the new 80 PLUS® Ruby certification, a standard introduced in January 2025 that sets the most stringent energy efficiency criteria currently applied to internal power supply products.
The power supply achieves a peak conversion efficiency of 97.5% when operated at 277 V input and 50% load, meeting the Ruby certification’s minimum requirement of 96.5%. The unit also reaches a power factor of 0.99 at full load, exceeding the Ruby standard’s minimum requirement of 0.96. These values indicate efficient transformation of electrical input into usable power, minimizing energy loss.
Designed specifically for high-power computing environments, the 5,500 W supply addresses growing demand in AI-based server infrastructure. It supports an input voltage range of 200V to 277 V and provides continuous output at 5,500 W. Compared to earlier-generation products, the new unit increases power density from 32W/in³ to 51W/in³, a 59% improvement, enabling higher output within the same rack space.
Delta’s soft-switching technology and advanced circuit topology at the power factor correction (PFC) stage enable this performance. Third-generation wide-bandgap semiconductor devices — such as silicon carbide (SiC) or gallium nitride (GaN) — have also been integrated into the power train. These components allow for high-frequency operation, which contributes to improved power conversion and thermal management.
The 80 PLUS Ruby standard is now the highest tier in the internationally recognized power supply efficiency certification system, having replaced Titanium. Ruby requires more stringent performance, especially at typical operating loads such as 50%, where AI servers frequently function. The certification is aligned with other international energy-efficiency programs such as ENERGY STAR and EU regulatory frameworks.
With increasing power demands from AI and data-centric workloads, redundant power solutions with this efficiency and power density offer options for optimizing rack space and reducing operational energy loss. This release reflects a shift toward high-efficiency components capable of sustaining intensive computational environments.
