While the world watched the World Cup, robots played their own version in Incheon. RoboCup 2026 ran from 30 June to 6 July in Songdo, drawing teams from dozens of countries. When the whistle blew, one name was behind every winner.

Teams using robots from Beijing’s Booster Robotics swept all three humanoid football divisions.

The numbers tell the story. Of the 59 teams in the humanoid leagues, 38 competed on Booster machines. They took gold, silver, and most of the podium across the Small, Middle, and Large divisions.

Tsinghua University’s Hephaestus team won the Large division on the Booster T1. Germany’s B-Human took the Middle division on the K1. A team called Invic won the Small division on the K1 Air.

From building robots to coding them

The clean sweep points to a real shift in the field. For years, each team built its own robot from scratch. Much of the work went into the mechanics, the hardware, and simply teaching the thing to walk.

That has changed. Most leading teams now buy the body off the shelf and pour their effort into software. They focus on perception, split-second decisions, and getting many robots to coordinate. Booster supplies the hardware and keeps improving the hard parts, such as running, sudden stops, and getting back up after a fall. The contest has moved from “who can build the robot” to “who can make it smarter.”

That split matters beyond football. Reliable legs and a stable body let researchers test complex “embodied intelligence” in the real world, not just in simulation. A robot that plays football is really a robot learning to see, balance, and react at speed.

The 2050 dream

The talk of beating human champions is not a fresh boast. It is RoboCup’s founding mission, set back in 1997. The pledge is that by 2050, a team of autonomous humanoid robots will beat the reigning World Cup winners under normal FIFA rules. “Our team’s ultimate goal is that we will beat the FIFA champion in 2050,” one competitor told Reuters in Incheon.

The gap is still huge, and the event was honest about it. RoboCup 2026 marked the first time two full teams of humanoid robots played 11 against 11 on real hardware. The scoreline was modest: B-Human beat fellow German side HTWK Robots 4:0. These are small, wobbly players, not Messi. Yet a decade ago even a steady walk was a struggle. Elsewhere, humanoids have already outrun a human record over a half-marathon.

China’s platform play

Booster’s win is also a business move. The company is not just selling robots, it is trying to own the platform they run on. It recently launched Booster Studio, which it bills as the first full development environment for embodied intelligence. Engineers use it to program, simulate, and deploy robot behaviour before touching real hardware.

One of the youngest teams in Incheon came from a Macau middle school. Its students trained their code in the simulator, then loaded it onto real robots. Booster has also started its own 3v3 robot football league to pull in more developers. The pitch is an open ecosystem, with Booster as the layer everyone builds on.

It fits a wider pattern. China shipped roughly 90% of the world’s humanoid robots last year, led by names like Unitree. The sector is crowded and not yet profitable for most, and Beijing has even started to register humanoids by ID. Winning a global contest is a cheap and vivid way to stand out.

Can they really do it?

A word of caution is fair. The flashy clips are heavily curated. A viral video showed a Booster T1 smashing a penalty clean through a wall, which is fun until you remember these machines share a pitch with people. Robots that kick that hard have already hurt bystanders at other events.

For now, robot football is a research tool dressed up as a spectacle. The gap to a real World Cup side remains enormous, and 2050 is a long way off. But the direction of travel is clear. The hard problem is no longer the body. It is the brain, and that is exactly the part improving fastest.