Humanoid robots are still limited by AI technology, according to Wang Xingxing, CEO of Hangzhou Unitree Technology.

Speaking at the World Robots Conference in Beijing, Wang said the industry may reach a breakthrough in one to three years, but current AI models are not yet advanced enough for mainstream humanoid robots.

Chinese robotics firms, including Unitree, have started trialing robots in factories for tasks like sorting and quality checks.

Large-scale deployment remains hindered by technical challenges. Unitree recently introduced its R1 humanoid robot, priced at 39,900 yuan (US$5,900), but it is not yet in mass production.

Other companies, such as Shenzhen-based Engine AI, are also planning to launch humanoid robots.

In 2025, Chinese humanoid robots joined events like half-marathons and football matches, but most struggled to complete the tasks, highlighting the technology’s current limitations.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ Chinese companies are pursuing aggressive pricing despite technological limitations

Unitree’s $5,900 price point represents a dramatic departure from industry expectations, showing how Chinese manufacturers are betting on volume over technological maturity.

Morgan Stanley projects humanoid robot prices will drop from around $200,000 in 2024 to approximately $50,000 by 2050 in high-income countries1. Yet Unitree and Engine AI are already pricing their robots at under $6,000, suggesting Chinese companies are prioritizing market entry over profit margins.

This pricing strategy aligns with China’s broader manufacturing approach. The country has led the world in humanoid robotics patent filings with 5,688 patents compared to the U.S.’s 1,4832.

The aggressive pricing occurs even as the CEO acknowledges that real-world applications remain vague, indicating Chinese companies may be positioning for future AI breakthroughs rather than current capabilities.

2️⃣ The humanoid robotics industry mirrors the pre-ChatGPT AI landscape

Wang’s comparison to the pre-ChatGPT moment reflects a broader pattern where hardware development often outpaces the software intelligence needed to make it truly useful.

The CEO’s prediction of a 1-3 year timeline for AI breakthroughs aligns with market forecasts showing slow adoption until the mid-2030s, followed by rapid acceleration in the late 2030s and 2040s1.

Recent Unitree developments