| Feng, Zhengyang, Jin, Tian and Wang, Sheng 2026. Power architecture evolution in data center. Presented at: Energy Conversion Congress & Expo Europe, Valencia, Spain, 14-18 September 2026. Proceedings of ECCE Europe. IEEE, 10.1109/ecceeurope67962.2026.11737924 |
Abstract
The explosion of generative artificial intelligence (AI) and large language models (LLMs) has catalyzed an unprecedented surge in compute power densities, with individual server racks now eclipsing 100 kW. This rapid expansion places severe thermal, spatial, and electrical strain on conventional data center power distribution systems and utility grids. Consequently, the industry is transitioning from passive consumption to an active, grid-interactive paradigm. This paper presents a comprehensive review of the evolution of data center power architectures, bridging the gap between macro-grid interconnection challenges and micro-scale semiconductor opportunities. We evaluate conventional low-voltage AC distribution, emerging high-voltage DC (HVDC) distribution, and solid-state transformer (SST)-based solutions. By comparing architectural trade-offs in efficiency, scalability, protection complexity, and technology readiness, this review identifies key technical challenges and outlines future opportunities for power electronics in next-generation data center infrastructure.
| Item Type: | Conference or Workshop Item - published (Paper) |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Engineering |
| Publisher: | IEEE |
| ISBN: | 9798319505651 |
| Last Modified: | 09 Nov 2026 13:52 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/190761 |
Actions (repository staff only)
![]() |
Edit Item |




Dimensions
Dimensions