Enteligent has launched a white paper detailing the adoption of 800VDC energy techniques for next-generation AI and GPU knowledge centres. The report highlights how conventional AC energy paths face financial and reliability challenges when supporting fashionable GPU clusters, whereas 800VDC techniques paired with 50VDC server architectures present a extra scalable and environment friendly answer.
Fewer conductors are wanted, with copper necessities lowered by 50–80%, based on the findings.
The paper notes that AI coaching clusters usually exceed 30 kW per rack, revealing the constraints of legacy AC techniques in thermal administration and energy supply. Excessive Voltage DC (HVDC) permits increased energy densities, supporting effectivity and reliability enhancements.
Adopting 800VDC techniques may end up in projected CapEx financial savings of $4 million to $8 million per 10 MW construct and an 8–12 % discount in annual energy-related OpEx.
The report additionally emphasises alignment with Open Compute Challenge requirements and the business shift towards DC-native architectures. Enteligent positions 800VDC as an acceptable basis for AI-first and hyperscale knowledge centres, supporting increased energy densities and long-term operational effectivity.
