China’s Robot Soldier Warning Emerges as Machines Break Olympic Records

Clear Facts

• A Chinese humanoid robot completed a 100-meter sprint in 9.32 seconds at Beijing’s World Humanoid Robot Games, faster than Usain Bolt’s 9.58-second human world record
• Chinese robotics engineer Yang Kun predicts humanoid robot soldiers could be deployed within five to ten years
• The U.S. Federal Communications Commission recently added foreign-produced advanced robotic devices to its Covered List, treating them as national-security concerns

A Chinese humanoid robot has shattered expectations by covering 100 meters faster than Usain Bolt’s 9.58-second human world record. At the opening of Beijing’s World Humanoid Robot Games, an X-Humanoid robot completed the 100-meter sprint in 9.39 seconds. Then the number dropped again.

X-Humanoid announced on Aug. 25 that TianGong Ultra ultimately posted a 9.32-second time during the competition. The company says its robots reset the mark several times during roughly two hours of racing.

Before anybody starts rewriting the Olympic record books, keep some perspective. Bolt still holds the official men’s 100-meter world record recognized by World Athletics. These robots compete under their own rules in a separate event.

Still, the speed at which this technology is advancing deserves serious attention. The bigger story goes far beyond a robot beating Bolt’s time. We are watching humanoid machines go from clumsy demonstrations to increasingly capable physical systems at remarkable speed.

The second World Humanoid Robot Games opened Aug. 22 at Beijing’s National Speed Skating Oval and ran through Aug. 26. Organizers registered 666 teams and 2,056 robots for this year’s event. The competition included 51 events and 1,301 competition sessions.

The robots compete in track and field events along with challenges designed around real-world jobs. Organizers have also added tests involving physical skills such as weightlifting and tug of war.

That mix reveals something important about what China wants these machines to become. The Games give developers a place to push speed and strength. They also expose weaknesses that can be harder to see in a carefully edited demonstration video.

The company says TianGong Ultra eventually reached 9.32 seconds, while other X-Humanoid teams recorded times of 9.34, 9.38 and 9.39 seconds. The exact comparison with Bolt needs context. These robots do not start, run or stop under the same conditions as human Olympic athletes.

However, that does not make the improvement meaningless. A humanoid staying upright at that speed requires serious coordination between its motors, balance system, sensors and software. All of those pieces have to work together while the machine continuously adjusts its body.

The rate of improvement becomes even clearer when you look back only one year. At the inaugural World Humanoid Robot Games in 2025, the winning 100-meter robot finished in 21.50 seconds.

Now robots from the same development ecosystem are running the distance in less than 10 seconds. That is an enormous leap in a short period. Earlier this year, a Chinese humanoid called Bolt reached 22 mph during testing.

Running fast looks impressive on video. Yet the technology underneath that speed could become useful in places far removed from a racetrack.

Running was only part of the show. An X-Humanoid robot reached 2.88 meters, or about 9 feet 5 inches, in a standing high jump. That height exceeds the recognized men’s high jump world record of 2.45 meters, or about 8 feet, set by Cuba’s Javier Sotomayor in 1993.

Again, we need context. Sotomayor’s world record came from a conventional high jump with a running approach. The robot competed in a standing high jump. Those events use very different mechanics. So the robot did not erase Sotomayor from the record books. What it did demonstrate was impressive explosive power and body control from a humanoid machine.

A fast humanoid robot does not need to become a track star to become valuable. Better mobility could help a robot move through a disaster area where sending a person could be dangerous. A machine that stays balanced while carrying equipment could eventually handle physically demanding jobs. Then you start thinking about environments where people would rather avoid going altogether.

That brings us to the part of this story that deserves much more attention. One developer at the Games believes humanoid robot soldiers could arrive within the next decade.

Robotics engineer Yang Kun, who works for Shanghai-based humanoid robotics company AG BOT, says that he believes Chinese companies could be only five to 10 years away from robot soldiers. His company’s A3 humanoid competed in kickboxing at the Games.

“Military and combat use as part of where I believe humanoid robotics could eventually go,” Yang described.

That prediction deserves context. This comes from one developer. It does not represent an announced Chinese government timetable for deploying humanoid soldiers.

Technology forecasts can move quickly in either direction. A machine can perform beautifully in a controlled environment and then struggle once the surroundings become unpredictable. What deserves attention is where developers believe this technology is headed. When people building these machines start discussing military applications, that tells us the conversation has already moved far beyond robots entertaining crowds at sporting events.

Running fast in a straight line is one challenge. Navigating a damaged building while carrying equipment presents an entirely different problem. A robot may encounter unstable ground, blocked paths and people moving unpredictably around it. Then comes the far harder issue of decision-making. A sprint gives a robot a clear destination. Real life rarely does.

Military environments make that problem even tougher because communication can disappear without warning. Sensors can misread a situation. Software may face circumstances developers never anticipated.

That is why perspective remains important when watching these record-setting performances. The robots are getting impressive. They also still fall down. At the Games, some fast-running robots had trouble stopping after crossing the finish line and crashed into padded barriers. That creates a visual reminder of the gap between raw speed and reliable control.

The organizers appear very aware of that gap. This year, Beijing upgraded the 100-meter event so robots must compete fully autonomously. Scenario-based events also encourage autonomous positioning, recognition and operation. The robots also face tasks inside settings modeled after factories, hotels and homes. Emergency-response challenges push them into situations where they must interact with their surroundings for longer periods.

A robot that runs 100 meters in nine seconds creates a great headline. A robot that can reliably work for hours without somebody constantly guiding it could change entire industries. China appears focused on both. Reports indicate efforts to scale humanoid manufacturing in China, with factories preparing to build thousands of machines.

That brings us to the larger technology race. We hear constantly about competition between the United States and China over artificial intelligence. Humanoid robotics adds a physical layer to that race.

The country that develops powerful artificial intelligence gains one advantage. The country that can put that intelligence inside a reliable robot body and manufacture those machines at scale could gain something much larger. Suddenly, the software can move through the physical world. It can enter an industrial facility. It can cross dangerous terrain or potentially operate where sending people would carry significant risk.

U.S. regulators are already treating advanced foreign robotics as a national-security concern. On July 28, the Federal Communications Commission added foreign-produced advanced robotic devices to its Covered List. The category includes qualifying humanoid robots and other advanced mobile robotic systems.

The change means covered new devices generally cannot receive the FCC equipment authorization required for import or sale in the United States unless they receive an applicable conditional approval. That tells you Washington sees connected robots as more than another gadget category.

For years, humanoid developers faced a frustrating mechanical challenge. How do you make a two-legged machine move efficiently without constantly falling over? Engineers still have plenty of work ahead. Yet today’s machines move very differently from the awkward humanoids we watched only a few years ago.

As robot bodies become faster and stronger, artificial intelligence continues improving at the same time. A more capable body could perform difficult physical work. Better AI could help that machine understand what it sees and determine what to do next. Put those advances together and humanoids become much more interesting.

The biggest hurdle now may become less about making a robot move and more about making sure it makes good decisions once it gets there.

You probably will not see a humanoid robot soldier walking down your street anytime soon. You may encounter the technology behind these machines much sooner. The balance system that keeps a robot upright during a sprint could help one move through rubble after an earthquake. More capable sensors could allow robots to operate around human workers more safely.

Factories will likely remain one of the first major targets. Most buildings and workplaces were designed around the human body. That makes humanoids appealing because companies could potentially deploy them without rebuilding every workspace around a specialized machine. As reliability improves and costs come down, those capabilities could expand into other areas.

The military discussion raises much harder questions. How much decision-making authority should an autonomous machine have? Who carries responsibility if an AI-controlled robot makes a dangerous mistake? Those conversations should happen before the machines become capable enough to force them. Because the technology is moving quickly.

The Bolt comparison grabs attention, but the bigger story is how quickly humanoid robots are becoming more capable. A 9.39-second 100-meter run shows how fast the motors, balance systems and software are improving. The military angle deserves attention too. One Chinese robotics engineer predicts robot soldiers could arrive within five to 10 years.

For America, this is another front in the technology race with China. A country that can combine advanced AI with reliable robot bodies and produce large numbers of them could gain a major advantage. We also need perspective. Running fast on a controlled track is far easier than navigating an unpredictable environment and making safe decisions. What deserves attention now is how quickly these stronger robot bodies advance alongside increasingly capable AI.

Let us know what you think, please share your thoughts in the comments below.