Beyond PUE: A Boundary-Aware Primary-Energy Framework for Evaluating Self- Powered Computing Infrastructure, with Application to a Hybrid Offshore Floating Data Center

Abstract Power usage effectiveness (PUE) is the dominant data-center sustainability metric, but it is defined only over the facility boundary: cooling and auxiliary overhead relative to the IT load. For self-powered infrastructure generating its own electricity on site, this boundary excludes exactly the losses—fuel-to-electricity conversion and any carbon-capture parasitic load—that determine whether total energy use is actually efficient. In this study, the hypothesis that facility-level PUE can systematically overestimate self-powered infrastructure sustainability is tested, and a boundary-aware primary-energy framework (η_primary = η_cycle × (1 − CCS penalty)/PUE) is developed to clarify that gap. The framework is applied to a proposal-stage architecture, the Hybrid Seabed-Energy Floating Data Center (HSE-FDC), offshore gas turbines, a wave-energy-converter array, carbon capture, and passive marine cooling, which had a PUE of approximately 1.04 with negligible grid dependency. Using literature-grounded parameters, the case study yields a plausible facility PUE of 1.08–1.09, which is comparable to that of real benchmarks, whereas the wave array contributes under 1.2% of the average power, which is far short of a genuine hybrid baseload. Critically, this favorable facility PUE corresponds to a primary energy efficiency of only 25.5–38.7%, below a conventional grid-tied facility (43.0%) and dramatically below a renewable-grid-tied facility (93.5%): a favorable facility PUE does not, by itself, indicate favorable total energy efficiency once onsite generation is included. The architecture's individual technologies are each already commercialized elsewhere; this study's contribution is the general framework and the demonstration that it changes the sustainability conclusion for a real architecture, generalizing to other off-grid contexts.