What is the power consumption of an industrial AI edge gateway under full load AI inference (in watts)?

The typical power consumption of an industrial AI edge gateway under full load for AI inference is 25W-45W, depending on the AI chip and peripherals. Higher power consumption increases power consumption and requires optimization of EMC. For filter inductor selection, we recommend using a low DC resistance common-mode inductor CML4532A-101T with a DCR of 18mΩ and a ferrite bead PBZ1608E600Z0T with a DCR of 0.05Ω to reduce heat loss. Simultaneously, the PCB layout should include heat dissipation vias and a copper power input, with a parallel TVS 5.0SMDJ33CA for surge protection and high-temperature resistance. Actual testing at 45W full load showed a power filter component temperature rise of <25℃, not affecting EMC performance.