While We Laugh at “Pointless” Humanoid Robots, China Is Already Running

A Robot Sports Meet in China

China recently held a sports competition for humanoid robots.

The robots ran, fell over, and competed against one another. The footage was quite entertaining.

In Japanese online discussions, however, I saw plenty of dismissive reactions:

“There is no point in making robots humanoid.”

“Wheels are more efficient.”

“Industrial robots are all factories need.”

“Have people forgotten the failure of Pepper?”

I understand the reasoning.

If all you need to do in a factory today is tighten screws, there is no reason to walk on two legs like a human. A fixed robotic arm will be faster, cheaper, and more precise.

If the question is which option makes more sense for immediate factory deployment, the industrial robot wins.

But this argument gives me an unpleasant sense of déjà vu.

Are we judging the value of the next technology only as an extension of what already exists?

The Reassurance That Japan Is Strong in Industrial Robots

Japan remains a major industrial robotics country.

It has globally competitive manufacturers such as FANUC, Yaskawa Electric, and Kawasaki Heavy Industries.

That leads to a reassuring interpretation:

“While China plays around by making humanoid robots run, Japan is earning money by building robots that are genuinely useful in factories.”

I understand the appeal of that view.

But we have heard this kind of reassurance many times before: Japan is strong in the current market, so there is nothing to worry about.

Being strong in today’s industrial robotics market does not guarantee strength in the next robotics market.

Why Make Robots Humanoid in the First Place?

The standard criticism of humanoid robots is that the human body is inefficient.

That is true if we look only at the robot.

If it only needs to carry cargo, wheels are enough. If it only needs to weld, a robotic arm is enough. Without stairs, there is little need for bipedal walking.

The picture changes, however, when we shift our attention from the robot to its environment.

Our society itself is designed around the human form.

The height of stairs, the position of door handles, the height of desks, the size of tools, and the width of corridors all assume a human body.

Traditional automation changes the environment to suit the robot. We build dedicated production lines, install safety fences, lay rails, fix components in specific positions, and redesign the work itself for automation.

Humanoid robotics aims to reverse that relationship.

Instead of adapting the environment to the robot, it adapts the robot to the human environment.

The robot can enter a workplace where a person was working yesterday. It can use human tools and move elsewhere when a different task is required.

If that becomes possible, simply comparing the efficiency of individual machines and saying “wheels are faster” no longer settles the question.

Should We Be Laughing at China’s Robot Games?

Depending on how you view it, the recent footage may look like nothing more than a robot festival.

But China is also manufacturing humanoid robots in large numbers, putting them into motion, allowing them to fail, and collecting the resulting data.

They fall. They fail to move properly. Then engineers fix them.

The cycle repeats at scale.

That is what technological development looks like.

“It is not useful yet” and “there is no point in researching it” are entirely different claims.

While people laugh at the current level of performance, China continues accumulating physical machines, data, and skilled people.

That concerns me far more than the sight of a robot falling over.

Pepper Can Bring the Conversation to a Halt

When humanoid robots are discussed in Japan, Pepper is almost certain to appear.

“Japan built Pepper years ago, and that failed.”

It is true that Pepper was not adopted as widely as many people expected.

But placing Pepper and today’s humanoid robots in the same category is too simplistic.

Pepper appeared before the generative AI revolution.

What is happening now is not merely progress in robotics. AI that perceives its surroundings, AI that understands natural language, AI that plans actions, and control technology that moves a body through the physical world are beginning to connect.

This is what people call physical AI.

The commercial struggles of one product called Pepper and the long-term potential of humanoid robotics as a technological category are completely different questions.

If an early product’s failure proved that an entire technology had no future, most of the technologies around us would have disappeared during their early stages.

Industrial Robot Manufacturers Are Providing the Answer

Do humanoid robots have a future? Will physical AI truly become the next enormous market?

Online commentators will not settle those questions by arguing with one another.

There is a much clearer answer.

Japan’s industrial robot manufacturers themselves have begun working with NVIDIA.

The people saying, “Japan already has world-class industrial robots, so there is no reason to pursue inefficient humanoids,” are outside observers.

The companies that actually build those industrial robots are already moving into the world of physical AI.

That makes sense when you think about it.

Those companies understand the strengths of industrial robots, the weaknesses of humanoids, and what is genuinely useful on a factory floor far better than online commentators do.

Yet they do not believe that continuing to refine conventional robots will be enough. They are working with NVIDIA and investing in the next generation of robotics.

We still do not know whether the eventual winner will be humanoid, wheeled, or a hybrid with robotic arms.

But at least one thing is clear:

The existence of industrial robots is not a reason to ignore physical AI.

After all, industrial robot manufacturers themselves are partnering with NVIDIA and beginning to pursue it.

That seems like the clearest answer of all.