For all the progress, most current humanoid deployments still require on-site engineering support from the robot's maker, custom preparation of the environment, and significant integration work. The idea of a general-purpose robot you can drop into any factory and switch on remains, for now, aspirational.
This is the sober counterweight to the sprinting-robot headlines: the machines work, but making them work at a given site is still a project, not a purchase.
What it means: businesses considering humanoids should budget for setup, integration and vendor support, not just the robot. The technology is real and improving, but the cost and effort of getting it running in your specific environment is the part that decides whether it pays off.