"What's the use of a robot running faster than Bolt?"

March 18, 2026

Word count: 2,072, estimated reading time: about 4 minutes

Author | First Financial Ning Jiayan

On March 17, the 26th Annual Yabuli China Entrepreneurs Forum opened. Outstanding entrepreneurs, experts, and scholars from various industries gathered under the theme “Confidence, Opportunities, New Chapter.”

A new board member and founder of Yushu Technology, Wang Xingxing, made a debut and spoke, with a remark “This year, robots will run faster than Bolt,” sparking widespread discussion. Amid the amazement, deeper questions arose: What does extreme speed really mean? When will we truly witness robots moving from “tech showcases” to “practical use”?

This delicate mindset, swinging between “technological amazement” and “lack of application scenarios,” actually reflects the market’s anxious anticipation of the “growth phase” of the robotics industry. Facing this heavy expectation, the industry needs not only to showcase strength but also to convey a clear development roadmap.

Heading into the “ChatGPT Moment”

The answer to “What’s the use of running fast” can be found in Wang Xingxing’s full speech. Reflecting on the company’s development, he mentioned that robots have made significant progress in AI, especially in integration and reinforcement learning. In 2023, robots could only walk simply; by 2024, they could perform relatively complex dance moves; and in 2025, under engineering mode, robots could execute various martial arts, even reaching levels difficult for ordinary humans to match.

This progress is driven by advancements in robot components, such as 3D laser radars improving localization in complex scenes, and algorithm improvements, like pre-trained IM models enabling flexible switching between actions, enhancing movement combinations, and developing full-body coordination and swarm positioning algorithms to ensure natural, seamless performance.

“Just like human athletes, as physical capabilities improve, their coordination and control also enhance,” Wang Xingxing believes. If a humanoid robot’s movement capabilities are rich enough to perform a variety of actions, its work efficiency will also be high. By continuously optimizing stability and consistency, robots can better adapt to large-scale applications across various scenarios.

“Robots completing 80% of tasks in 80% of unfamiliar scenarios—that’s the ‘ChatGPT moment,’” Wang Xingxing states. He believes the industry still needs to solve three key issues: first, improving models’ ability to express tasks and actions, breaking through generalization bottlenecks; second, increasing efficiency in utilizing multi-modal data like videos, simulations, and real machines, reducing reliance on large-scale real-world data collection; third, enabling reinforcement learning to form reusable, scalable effects. Regarding technical routes, he favors world models and video generation models, believing these paths have higher ceilings and more potential to leverage the vast amount of internet video and text data.

Mass Production Is the Key

What are robot companies busy with this year? Mass production is the keyword.

Songyan Power, originally scheduled to attend the Yabuli Forum, was unable to attend due to other commitments. Founder Jiang Zeyuan told First Financial that Songyan Power has no new product launches planned in the short term. Their goal this year is to sell over 10,000 units, a target set for the 2026 growth phase.

In an interview with First Financial and others, Jiang Zeyuan mentioned that humanoid robots consist of hundreds or thousands of small parts. If any component runs out of stock, production stalls. Additionally, increasing monthly production from hundreds to thousands units requires a large workforce, posing management challenges. This year’s delivery goal is to achieve breakthroughs in both supply chain and manufacturing.

He believes the industry has entered a preliminary scale-up stage this year. While robots are not yet common in homes for laundry or cooking, total robot sales are expected to significantly increase. “We hope to become the top player in a specific industry or scenario within about three years, capturing over 60% market share, though the exact industry or scenario is still uncertain because the industry changes too fast. Three years ago, I couldn’t have imagined today’s Songyan Power,” he said.

“Reaching the scale of ten thousand units is a major milestone for humanoid robots. Songyan Power emphasizes delivering products to end households and achieving real application in scenarios,” Jiang Zeyuan added.

2026 will surely be a year of intensified Matthew Effect in embodied intelligence, with leading companies growing stronger. Companies reaching mass delivery or making breakthroughs in core technology will further accelerate this effect. This year is expected to be a year of continuous industry breakthroughs.

Full Industry Support and Integrated Efforts

To realize the “ChatGPT moment” for robots, the industry is working together: improving supply chains for manufacturing, promoting accessible and affordable applications, and establishing a closed-loop after-sales system to ensure comprehensive implementation.

JD.com officially launched the “Smart Robot Industry Acceleration 2.0 Plan” at the 2026 China Home Appliances and Consumer Electronics Expo. JD’s smart robot division head stated that JD will invest hundreds of billions of yuan into smart robots, promoting the formulation and improvement of industry standards.

To address issues like short battery life and poor adaptability, JD has introduced standardized robot battery solutions. These solutions have already been validated with multiple robot companies.

In the leasing sector, CEO Li Yiyan of Qingtian Leasing told First Financial that, based on current market conditions, most robot leasing clients are B-end, including brand events and shopping mall traffic. These clients prefer “quick deployment, low-threshold trial, and scenario-based payment,” which are the main sources of current orders. The C-end market has also begun to emerge but remains in early stages, mainly for light, trial-oriented needs like family gatherings and weddings, characterized by strong social and viral attributes, making it easier to build user awareness and market education. “We believe that in the near future, the robot leasing market will show a clear trend: B-end will first establish business models, and C-end will gradually develop consumer habits, jointly expanding industry scale,” he said.

“China’s robot market is already among the world’s leading, and global industry players are optimistic about China’s prospects. Ultimately, the competition in the robotics industry is about implementation capability rather than just display effects,” said Sun Rongyi, Vice President and Director of Qianxun Intelligence.