Booster Robotics wants to turn kids into humanoid robot developers

Booster Robotics wants to turn kids into humanoid robot developers

John KoetsierNovember 17, 2026
ShareXLinkedInFacebook
China’s Booster Robotics is betting that tomorrow’s humanoid robot developers are today’s schoolchildren. With 3,000+ robots delivered across 30+ countries and hardware powering nearly two-thirds of RoboCup 2026 teams, Booster is building an open platform for physical AI, hoping developers will bring their own intelligence to its robots.

From soccer to school

At RoboCup 2026, 38 of the 59 competing humanoid robot soccer teams used hardware from one company: China's Booster Robotics. And teams using Booster robots won the championship in all three humanoid divisions.

That's not a bad result for a company founded just three years ago.

But Booster has a bigger ambition than winning robot soccer games. It wants to become one of the standard platforms that developers use to build physical AI applications, and it's starting with schools, universities, and research labs.

"The main use case is for education, for primary or middle schools," Booster's Clarence Huang told me at Web Summit in Lisbon, describing the company's smaller K1 humanoid. "So that the young students can have the basic knowledge of physical AI."

And that could be an interesting long-term strategy.

The next generation of robot developers

Booster Robotics is a relatively young company, founded in 2023, but its technology comes out of Tsinghua University's robotics research program.

"It's a young company, but the technology starts from 20 years ago," Huang told me.

The company's K1 humanoid is a small robot, just 95 centimeters tall and weighing 19.5 kilograms. It has 22 degrees of freedom, a depth camera, and onboard AI computing power ranging from 48 to 200 TOPS depending on the configuration.

In other words, it's a real programmable humanoid robot, but small enough for students to work with in a classroom or robotics lab.

"20 or 30 years ago, I learned the computer in school," Huang said. "It's the same thing nowadays, but with robots."

That's an interesting comparison. Personal computers became ubiquitous in part because millions of children learned to use them in school. Booster is betting physical AI will follow a similar path.

And the company is putting some resources behind the idea.

In February, Booster supported an inaugural secondary-school humanoid robot soccer tournament, providing curriculum development, training, and technical support. It has also launched an initiative targeting 10,000 schools across 100 cities.

There's a commercial strategy here as well.

A student who learns to program a Booster K1 today might be a robotics researcher or startup founder in five or ten years. And when that developer needs a humanoid platform for a new project, Booster hopes its robots will already be familiar.

Robot soccer is becoming a software competition

The RoboCup results show how this approach might work.

At the 2026 championship in South Korea, teams using Booster's T1 won the Large division, while K1-powered teams won the Middle division and K1 Air robots took the Small division championship.

Booster hardware appeared in nearly two-thirds of the competing teams. That's significant because robot soccer is increasingly about the software running the robots, rather than the mechanical design of the robots themselves. When multiple teams use similar hardware, the competition becomes a test of perception, movement planning, coordination, and AI.

Booster also makes its hardware more accessible to competing teams. Under the RoboCup Humanoid Robot Program, the K1 RoboCup Edition has a listed price of $14,999, discounted to $9,999 for eligible teams. The idea is to get more developers building on the platform, even if that means making less money on an individual robot.

You bring the intelligence

Booster takes a somewhat unusual approach to humanoid robotics.

"Actually, Booster, we produce the hardware," Huang told me. "But we do not focus on the software."

That's not entirely literal. Booster develops operating systems, development tools, and some AI technology. But the larger strategy is to provide hardware that other companies can program with their own robot intelligence.

"You can come in and put your own foundation model," Huang said. "You can do whatever you want."

The company also provides sample code, APIs, and software development kits so developers don't have to start from scratch. And Booster Studio, its integrated development environment, attempts to make the entire process easier.

"Nowadays we know that the developers, they use one app for coding, another app for debugging, another app for sim to real," Huang said. "But we combine all in one platform."

That's potentially useful because moving from simulation to a physical robot remains a significant challenge in robotics. Booster's approach is to provide the tools for building, testing, and deploying robot behaviors in one environment.

The platform is free to use, Huang says.

Building robots for other people's AI

Booster's larger T2 humanoid is designed for more sophisticated research and potential industrial applications. It stands 1.4 meters tall, weighs approximately 42 kilograms, and can carry up to 10 kilograms across both arms. The Professional version runs on NVIDIA's Jetson Thor T5000, with a theoretical peak of 2,070 TFLOPS of AI computing power.

But the hardware is configurable.

Huang told me Booster doesn't manufacture its own dexterous hands. Instead, it integrates hands from third-party suppliers, including BrainCo and Sharpa.

That allows developers to select the hands that suit their particular applications, rather than being locked into a single design.

There's another interesting feature: magnetic charging while operating. At Web Summit, Huang showed me a T2 running while connected to its charging system.

"It can work during its charging," he said.

Without that connection, the robot has approximately two hours of continuous walking endurance. With power connected, it could potentially perform stationary tasks for much longer, although the software and application would still determine how autonomous it could be.

But the question for industrial deployment isn't just hardware capability.

It's safety.

When I asked Huang whether the robot was safe to operate alongside humans, he didn't make unrealistic promises.

"We cannot say it's 100% safe," he said, adding that safety depends on the maturity of the software, the complexity of the task, and the environment.

That's especially relevant when the customer is supplying the robot's intelligence.

Can Booster turn developers into customers?

Booster says it has now delivered more than 3,000 robots across more than 30 countries and regions. It also raised nearly $150 million in April to expand its delivery network, research, and product development.

Those numbers suggest its education-and-development-first approach has real traction.

The next challenge is converting that ecosystem into industrial applications. After all, most factory operators don't want to build their own robot AI. They want robots that perform useful tasks with minimal development work.

But Booster doesn't necessarily need to sell directly to those customers. It can sell to AI startups, systems integrators, and other robotics companies that turn its hardware into finished products. And that's where the company's bet on education could become particularly valuable.

If the next generation of robotics developers learns physical AI on Booster hardware, the company could become an important platform supplier without ever needing to build the world's best robot foundation model.

When I asked Huang what was coming next, he wasn't ready to announce a specific new product.

"Maybe we will have other models," he said. "But we will focus on the hardware," and bringing more developers onto the platform.

There's no guarantee that today's classroom robot will become tomorrow's industrial standard. But there's a reason technology companies have spent decades trying to get their products into schools.

Sometimes the most important customer isn't the one buying your technology today. It's the one learning to use it.

Share this article

ShareXLinkedInFacebook

Written by

John Koetsier
John Koetsier

I tell stories with data and build startups. I've raised angel and VC funds, led software companies and startups, and chronicled the rise of the mobile economy. Currently I write for Forbes, host several podcasts, and am starting up HumanoidMegaHub. I was TUNE's mobile economist and Singular's VP of Insights, built the research team at VentureBeat, and managed teams creating software for partners like Intel and Disney. In past lives I built websites, e-learning platforms, and mobile apps, and, a long time ago, was named to Folio's top 100 of the media industry's "most innovative entrepreneurs and market shaker-uppers.”