ROBBRIEF
The Robot Report argues the binding constraint on factory humanoids is the cost stack, not AI capability, and a modeled 1 human per 50 robots supervisor ratio is the mechanism that has to close.
The Robot Report's August column, "Humanoids won't scale on factory floors until costs drop," argues physical AI has caught up. Actuators are stronger, mobility is better, and hands are "close enough" for many factory tasks. The column names Generalist, Agility Robotics, and Alt-Bionics as vendors with operational results. The question that decides whether humanoids scale is no longer capability; it is how many robots one human can supervise before the per-unit cost stack collapses.
The column's proposed mechanism is a 1-to-50 ratio: one operator steering 50 or more robots, intervening only on exceptions, with each intervention feeding the training loop. The column presents the number as a model rather than a measured figure, and the interventions never reach zero.
The falsifier is older than the hype: purpose-built robots still beat humanoids on per-task economics. Humanoids win on flexibility (one machine, many jobs, minimal reconfiguration). Until the supervisor ratio moves from model to measurement, the invoice will keep winning.