This year's humanoid robot trend has been very intense. Capital continuously flows into startups, and investors are full of confidence—this thing will soon be in warehouses, on assembly lines, and even entering ordinary households. It sounds fantastic, but problems also arise.

People outside the industry worry that this is a bubble, and more noticeably, some players inside the industry have started to hit the brakes. The reason is quite simple: there are too many engineering challenges and commercial costs between laboratory demonstrations and products that can truly replace human labor.

The most convincing voices come from companies that have already secured actual orders. Agility Robotics, which makes the bipedal robot Digit, now has hundreds of machines working in real scenarios—sorting and搬运 in Amazon warehouses, and appearing on production lines of automotive parts manufacturer Schaeffler. Sounds impressive, right?

But even with these achievements, Agility’s CTO Pras Velagapudi still poured cold water: don’t be fooled by the claims that the "humanoid robot wave is about to fully arrive." He said very straightforwardly—making robots move boxes in relatively clean warehouses is one thing; turning them into intelligent housekeepers capable of handling various complex situations at home is a completely different matter. Engineering complexity, stability, costs... each is a hurdle.