Who Will Lead the Race to Build the First True Humanoid Robot?

The Global Race to Build the First Practical Humanoid Robot

On a crisp spring morning in Hanover, Germany, at the massive Hannover Messe industrial trade show, all eyes are on a small but remarkably agile humanoid robot—the G1, developed by Chinese robotics company Unitree.

Standing at just 4'3" (130cm), the G1 is smaller and more affordable than many of its competitors, yet its dexterity and lifelike movement have turned it into an internet sensation. Videos of it dancing and performing martial arts have captured widespread attention. At the expo, it’s being remotely controlled by Pedro Zheng, Unitree’s sales manager. He explains that while G1 can be programmed for autonomous functions, current users must configure them individually.

The G1’s human-like shape attracts a steady stream of curious onlookers. They interact with it instinctively—offering handshakes, laughing at its gestures, even apologizing when they bump into it. While plenty of machines are on display at the show, few capture the public's imagination quite like this humanoid figure.

Unitree is one of dozens of companies worldwide racing to build robots in the image of humans. The commercial potential is enormous: humanoid robots could revolutionize everything from manufacturing and logistics to domestic chores, becoming tireless workers that don't need rest or pay raises. Yet, major technological challenges remain before such visions become reality.

Traditional industrial robots have operated in factories for decades, where controlled environments minimize risk. Introducing a humanoid robot into unpredictable spaces—like homes or restaurants—poses safety and complexity challenges. Strength is necessary for usefulness, but a falling robot could be dangerous. And perhaps the biggest hurdle is the artificial intelligence needed to make humanoid robots truly autonomous.

“The AI simply hasn’t reached a breakthrough moment,” says a Unitree spokesperson. Current robot AI struggles with basic reasoning and logic, making it difficult to handle complex tasks without human input. As a result, robots like the G1 are mainly sold to research institutions and tech firms for development purposes.

For now, the focus is on using humanoid robots in factories and warehouses. Elon Musk’s Tesla is developing the Optimus robot, with "several thousand" units planned for 2024 to perform practical tasks in Tesla factories. Other automakers are following suit—BMW has introduced humanoid robots in a U.S. plant, and Hyundai has placed massive orders with Boston Dynamics, which it acquired in 2021.

According to Thomas Andersson, founder of robotics research firm STIQ, 49 companies are currently developing humanoid robots with two arms and legs. If the definition is broadened to include robots with wheeled mobility, the number exceeds 100. He believes Chinese firms will dominate, thanks to strong supply chains, a robust R&D ecosystem, and government support. In fact, nearly 60% of humanoid robot investment is concentrated in Asia.

Unitree’s G1 illustrates China’s advantage, offering a humanoid robot at a competitive price of $16,000 (£12,500). Meanwhile, European developers face logistical and cost-related hurdles. Bristol-based entrepreneur Bren Pierce, founder of Kinisi Robotics, highlights the issue: most components are sourced from China, making it inefficient to manufacture elsewhere. His company’s KR1 robot, designed for industrial use, skips humanoid legs in favor of wheels, which are cheaper and more practical for warehouse floors.

“We use as many off-the-shelf parts as possible—motors, batteries, cameras—everything is mass-produced,” says Pierce. His goal is usability. Unlike complex robots that require expert knowledge, the KR1 is built so an average worker can learn to use it in a few hours. After 20 to 30 demonstrations, the robot can replicate a task independently.

The KR1 is being piloted by select customers this year. As for robots entering homes anytime soon? Even Pierce, a self-described optimist, sees that vision as at least a decade away.

“My dream has always been the ‘everything robot’—a general-purpose home assistant,” he says. “But it’s an incredibly complex challenge. I still believe we’ll get there, but not for another 10 to 15 years.”