Power Usage Effectiveness (PUE)
DC-01Total facility power against IT load — the headline efficiency ratio, and the reporting boundaries that decide whether it means anything.
energy curvatureefficiency metricsPUEenergyefficiencypower
Data Centre Datasets — Energy, Cooling & Efficiency
Authoritative excerpts drawn from the IEA Electricity 2024 report, Uptime Institute Global Data Center Survey, LBNL data-centre reports, and hyperscaler sustainability disclosures. These datasets provide natural rods for understanding energy use, thermal load, water consumption, grid carbon intensity, and the compute footprint of AI training.
Dataset family
Water Usage Effectiveness (WUE)
DC-02Litres of water per kilowatt-hour of IT load, including the off-site water embedded in the electricity a site consumes.
cooling arcsefficiency metricsWUEwatercoolingconsumption
Thermal load and rack density
DC-03Rack density, inlet temperature envelopes, and the shift from air to liquid cooling as heat flux per rack rises.
cooling arcsenergy curvaturethermalrackdensityairflow
Grid carbon intensity of compute
DC-04Carbon intensity of the grid serving the load, hour by hour — the difference between nominal and actual emissions of a compute job.
energy curvatureefficiency metricsgridcarbon-intensitycomputeemissions
Network latency and edge distribution
DC-05Latency budgets and edge placement: how geographic distribution trades transmission cost against local efficiency.
efficiency metricsenergy curvaturelatencyedgethroughputtopology
Storage lifecycle and e-waste
DC-06Drive and server lifecycle, embodied carbon, refresh cadence, and the e-waste stream that follows a hardware generation.
efficiency metricscooling arcsstoragelifecyclee-wastehardware
Workload placement and utilisation
DC-07Server utilisation, idle draw, and scheduling — the gap between provisioned capacity and work actually performed.
efficiency metricsenergy curvatureworkloadutilisationschedulingcapacity
AI training energy demand
DC-08Energy demand of frontier model training and inference, and the growth curve that now drives new capacity planning.
energy curvaturecooling arcstrainingGPUenergydemand
Rod-set downloads
Each category resolves to a JSON rod-set the engines can read directly.