An interactive model comparing effective synaptic operations per second deliverable at a fixed wall-plug power by superconducting optoelectronic networks versus conventional GPU farms. Sliders set the total power budget, the SOEN device energy per synaptic operation, the 4 K cooling overhead, and the GPU effective energy per synaptic operation. Outputs show synaptic-ops-per-second for each on a log scale, the ratio, human-brain equivalents, and how the SOEN power budget splits between compute, cooling, and support.

1.00 GW
20 aJ
1000×
2.0 pJ
Synaptic operations per second (log scale, 10¹⁸–10²⁴)
SOEN
GPU farm
Efficiency edge
SOEN ≈ brains
GPU ≈ brains
Where the SOEN power budget goes
compute cooling support