The robot completed 21 km. That alone is enough to change the premise.

On April 19, 2025, at the Beijing Half Marathon, the humanoid robot “GOFA” sent by a Chinese robotics company completed the 21.0975 km course in approximately 2 hours and 29 minutes. While this is far from the top human runners (the world record is 57 minutes and 31 seconds), those who laugh and say, “The robot lost to humans” are missing the essence of the matter.

The point is this: A bipedal humanoid robot autonomously completed a public road half marathon without falling or running out of battery. This was not possible just a year ago.

When measuring the evolution of technology, one should look at the “difference from last year” rather than the “absolute performance of today.” As of 2023, humanoid robots making headlines for walking just a few hundred meters outdoors have now progressed to completing 21 km in just 1-2 years. This acceleration is crucial information that directly impacts management decisions.

The Expiration Date of “Robots Are Still Not Usable”

When talking to managers of small and medium-sized enterprises in rural areas, the conversation often turns to robots and automation, and they frequently respond:

“Our work environment is complex, so robots are still not feasible for us.”

This judgment may be correct as of today. The problem is, how long will that “correctness” last?

Let’s look at some numbers:

• 2022: Boston Dynamics’ “Atlas” demonstrated carrying tools within a factory. The price is undisclosed but estimated to be in the millions to tens of millions of yen per unit.
• 2024: China’s Unitree’s versatile humanoid robot “G1” began pre-orders at approximately 2.5 million yen.
• 2025: Multiple humanoid robots participated in the Beijing Half Marathon, with several completing the race.

In just two years, prices have dropped by more than one digit, and performance has increased several times over. This follows the same curve as past drones and smartphones. Drones, initially military-grade and costing millions, can now be purchased for hundreds of thousands of yen from DJI for industrial use. They have seamlessly integrated into rural areas for pesticide spraying, surveying, and inspections.

Humanoid robots are likely to follow the same path. The time it takes for “still not feasible” to change to “too late to catch up” is shorter than many people imagine.

The Key Point Is Not That It “Ran” but That It “Did Not Fall”

The most technically significant aspect of the Beijing Half Marathon is not the time. It is the fact that the robot did not fall while autonomously navigating 21 km of public roads with uneven surfaces, inclines, and mixed environments with other runners.

This indicates that control technology, sensing, battery management, and mechanical design have all surpassed a certain level. According to reports, some of the participating robots were equipped with liquid cooling systems, addressing heat issues during prolonged operation. This technology is said to be adapted from smartphone gaming cooling techniques.

This “cross-domain technology adaptation” is effective. Not only robotics manufacturers but also smartphone manufacturers (like Honor) and home appliance manufacturers are entering the field because existing mass production technologies—batteries, motors, sensors, cooling—can be directly utilized.

The ability to use mass production technology means that costs will drop rapidly. This will be the most significant structural change.

Will “Robots” Be Included in Labor Cost Calculations Three Years from Now?

Let’s think concretely.

In a rural manufacturing business, suppose three part-time workers are hired for picking and transporting within a warehouse. They earn 1,100 yen per hour, work 8 hours a day, and operate 22 days a month. Each worker costs about 190,000 yen per month, totaling 570,000 yen for three workers, or 6.84 million yen annually.

Now, imagine that in three years, a humanoid robot capable of performing similar tasks can be purchased for 5 million yen, with a monthly running cost (electricity and maintenance) of 30,000 yen. With two units, the initial cost would be 10 million yen, and the annual running cost would be 720,000 yen. This would break even in the second year.

Of course, this price and performance are not yet realized as of today. However, considering the fact that the Unitree G1 is available for 2.5 million yen and the annual performance improvement curve, the idea of a robot capable of warehouse tasks for 5 million yen in three years is not far-fetched.

The question is whether to exclude this possibility as something “still far off” from management plans or to incorporate it as an “emerging scenario.”

The Real Risk for Small and Medium-Sized Enterprises

Large corporations are investing in robotics research and development, advancing pilot implementations, and accumulating know-how. Tesla’s “Optimus,” Figure AI’s “Figure 02,” and Chinese companies like Unitree and UBTECH are seriously investing in the practical application of humanoid robots.

On the other hand, what about small and medium-sized enterprises in rural areas? In many cases, the moment they hear “robot,” their thinking stops at “this has nothing to do with us.”

However, looking back at history, the democratization of technology has always flowed from “large corporations to small and medium-sized enterprises.” The same was true for the cloud. What used to cost hundreds of thousands of yen for in-house servers has now become a few tens of thousands of yen with AWS. The same goes for accounting software. Monthly financial statements that were previously handed off to accountants have been automated with freee.

Robots will follow the same trajectory. Initially, they will only be found in large corporate factories. But in 3-5 years, they will drop into a price range that small and medium-sized enterprises can “buy and use.”

At that time, there will be a significant difference in the speed of implementation between “companies that begin considering it for the first time” and “companies that already have an automation design philosophy.”

This is the real risk. It is not the risk of not being able to buy a robot. The risk lies in being unable to act the moment robots become affordable.

What Can Be Done Starting Today

This is not a call to buy robots immediately. What needs to be done today is more modest and more fundamental.

1. Inventory the “tasks performed by humans” in your company.

Identify which tasks require how many people and how many hours. Calculate the hourly wage. Just listing this information is sufficient. Surprisingly, many small and medium-sized enterprises do not have this basic data.

2. Reassess with the perspective of “Could this be done by machines?”

It is not necessary to automate immediately. Just categorize tasks as “potentially replaceable if technology catches up.” This will change how you view technology news.

3. Start with small automation.

You don’t need humanoid robots right away. Automate invoice processing with RPA. Reduce inquiries with AI chatbots. Use collaborative robots (cobots) for simple transportation tasks. These experiences of “small automation” will lay the groundwork for future large-scale automation.

The Speed of Updating What “Robots Can Do”

Finally, I want to emphasize one last point.

The essence of the news about the humanoid robot completing a half marathon is not that “robots are amazing.” It is that the speed of updating what “robots can do” is exceeding the expectations of many managers.

What was impossible a year ago is now possible today. What is not possible today may become possible in a year. This cycle is accelerating.

Management decisions are typically made on an annual basis. Medium-term management plans span 3-5 years. However, the evolution of robotics is rewriting what is possible on a monthly basis.

The cycle of management decisions is no longer keeping pace with the cycle of technological evolution.

That is why what is needed now is not a decision on whether to implement robots, but rather a habit of continuously observing the speed of technological evolution and aligning it with your company’s operations.

The robot that completed the marathon in 2 hours and 29 minutes has not yet arrived at your company’s warehouse. But are you prepared for its arrival?

Creating a state where you can answer that question is the one thing you should do after reading this article today.