OCP accepts Airsys whitepaper on data centre power allocation
Author: Joe Peck
The Open Compute Project (OCP) has accepted a whitepaper by Airsys executives proposing a new metric to measure how much of a data centre’s provisioned electrical capacity is available for IT compute before construction begins.
The paper introduces ‘Power Compute Effectiveness’ (PCE), a metric intended to help data centre designers identify how much power is allocated to IT equipment, cooling systems, electrical losses, and auxiliary loads at the design stage.
The framework was authored by Airsys founder and CEO Yunshui Chen, Chief Strategic Relations Officer Paul Quigley, and Product Marketing Manager Jake Roberts. It was contributed to OCP, peer reviewed, and accepted under an open licence.
As AI workloads increase power density and grid connection constraints limit available capacity, the paper argues that operators need to understand how electrical capacity is allocated before a facility is built.
Existing metrics, such as Power Usage Effectiveness (PUE), assess operational energy efficiency rather than how provisioned capacity is distributed in the design.
PCE measures capacity available for compute
PCE measures the proportion of provisioned electrical capacity within a defined facility boundary that is available for IT compute after accounting for redundancy requirements. Expressed as a value between zero and one, it is intended to complement established operational metrics.
The proposed methodology has three main applications: identifying capacity allocated to cooling, electrical distribution losses, and auxiliary loads; complementing PUE by assessing capacity allocation at the design stage; and enabling comparisons across facilities of different sizes, including colocation sites, hyperscale campuses, and retrofit projects.
Yunshui Chen says, “Operators have spent years optimising for how power is used inside a data centre using existing metrics like PUE. PCE asks a different question: how much of the power facility you’ve already paid for is structurally reaching compute in the first place?
“We built this framework to give the industry a common, design-level way to answer that question, and we believe the answer should benefit every operator, not just Airsys customers.”
Paul Quigley adds, “PUE has served this industry well, but it looks in the rearview mirror at operational energy efficiency. PCE is meant to be the windshield.
“A design team can take the facility design documents, starting with the single-line diagram, and determine how much of the provisioned power is allocated to compute and how much is allocated to cooling, electrical losses, and auxiliary load, before a single watt is drawn.
“When power decides how big a data centre can be, that’s a question every operator should be able to answer on the drawing board.”
Whitepaper sets out design-stage approach
The paper presents PCE as a provisional metric that could help operators, owners, and planners assess capacity allocation before construction, or when evaluating facility upgrades and retrofits.
Jake Roberts notes, “Energy stewardship starts long before a facility goes live. The path every kilowatt takes between the utility meter and the rack is more nuanced than most people realise.
“PCE is a provisional metric, and the earlier we can calculate the power allocation for a facility – starting in the design phase or in the retrofit/upgrade stage – the better we can steward that power for its entire life.
“Having OCP peer review and accept the whitepaper opens the door to broader industry adoption, and I’m excited for a future where we can plan with metrics that help us prioritise power allocation and stewardship from the start.”
The whitepaper, Power Compute Effectiveness (PCE): A Provisioned Capacity Allocation Metric for the New Data Center Era, is available through the OCP Contribution Database.
For more from Airsys, click here.
