humanoid robotics, which Tan said has the ability to redefine manufacturing
Redefine into shit? Manufacturing is the area where humanoid robots make the LEAST sense. Humanoid robots are useful when you need a robot to operate in an existing environment designed for humans. In a factory, you can control the entire environment. Application-specific automation will always beat humanoid robots.
The only caveat is if they somehow became cheap enough because you can do the design once and stick them everywhere, whereas a task-specific machine needs to be designed per-task. I suppose they're more interchangeable as well, but they're still shit for scale.
Redefine into shit? Manufacturing is the area where humanoid robots make the LEAST sense. Humanoid robots are useful when you need a robot to operate in an existing environment designed for humans. In a factory, you can control the entire environment. Application-specific automation will always beat humanoid robots.
The only caveat is if they somehow became cheap enough because you can do the design once and stick them everywhere, whereas a task-specific machine needs to be designed per-task. I suppose they're more interchangeable as well, but they're still shit for scale.
A humanoid robot can be trained just by watching a person do the job, like Google's AI robot (although most of their PR video was faked).
That takes all the cost out of it other than the robot.
The robots aren't up to it yet, but that's the theory anyway.
Compare this: https://www.youtube.com/watch?v=AL4mbWb2NkI?t=20
With this: https://www.youtube.com/watch?v=ssZ_8cqfBlE
Building new workflows around purpose-built robots is so much more efficient.