Training nodes trip municipal PDU budgets under peak GPU draw. What UCS control should the city adjust?

When training nodes trip municipal PDU budgets under peak GPU draw, the city should adjust the UCS power policy so rack power draw stays within measured PDU budgets. Power policy is the UCS control-plane lever to constrain or prioritize server power when facility capacity is the binding constraint on the AI rack. Domain 3.2.b frames that response explicitly for civic GPU deployments facing breaker trips. Adjusting the UCS power policy to raise caps for sustained peak GPU training throughput allows even higher draw and will keep tripping the measured PDU budget under peak load. Adjusting the UCS power policy for grid redundancy across measured dual power feeds addresses power source resilience but does not cap rack power draw and cannot stop PDU trips under peak GPU load. Adjusting the UCS power policy to sequence blade discovery after utility power restoration staggers boot after an outage but does not constrain sustained peak training draw and will not prevent breaker trips during workloads. Exam caveat: after capping power, training throughput may drop—ops must balance job SLAs against electrical reality rather than deleting alarms. Operational check: capture PDU trip and peak-draw data for the training racks, set UCS power policy limits aligned to the measured budget, re-run peak GPU workloads, confirm trips cease while Intersight still reports power telemetry, and record the new envelope for municipal capacity planning.