[Bloomberg Column] Humanoid Robots Still Have a Long Road to Commercialization

Investor Hype Far Outpaces Reality
Calls for a "Healthy Dose of Skepticism"

The event that captured international attention was when Unitree Robotics, a Chinese robotics company, was listed on the Shanghai Stock Exchange on August 19 and its share price surged up to 629% above its public offering price. As both investment banks at home and abroad, as well as individual investors, expressed high expectations for the future of the robotics industry and jumped into investing, robot-related stocks in other countries also experienced upward trends.

Yet, Wang Xingxing, CEO of Unitree, expressed strong doubts about the future of the robotics industry. At the World Robot Expo held in Beijing on August 20, he pointed out, "Robots perform impressively in controlled environments. However, they are still far from being established as general-purpose technologies in real life." He added, "It will take at least another 10 years before robots can enter mass markets," and even called that outlook quite optimistic.

While investors talk only about a rosy future, professionals actually working in the sector argue that a healthy dose of skepticism is needed regarding the humanoid robot boom. The dazzling demonstrations and viral promotional events have fueled an illusion that technological progress is moving very quickly, but in reality, robots still face many technical limitations before they can be used in practical applications.

Not only CEO Wang, but also other robotics experts agree that for humanoid robots to become commercial products, they first require the brains to be installed in their bodies. While hardware has advanced sufficiently, the AI systems required to move the mechanical bodies remain in their early stages of development. Even the best humanoid AI systems developed to date perform at only about 10% of the level at which the human brain controls the body. Developing AI that can freely operate a robotic physique is a task on a completely different level from building large language models (LLMs) like the currently commercialized ChatGPT.

Nonetheless, in the minds of investors, humanoid robots are already seen as innovative household appliances that will soon be introduced into every home, replacing domestic chores. The overheating in the market is evident from the subscription scale for Unitree's IPO being 5,500 times the offering. Individual investors and corporate investment institutions alike have been thoroughly captivated by robots showcased at events such as the World Humanoid Robot Olympics and by dancing and kung fu-performing robots on television.

In particular, unlike many other technologies, humanoid robots are expected to lead trends in the field of technology for a long time to come. Each time one of Unitree’s marketing videos is released, it records hundreds of millions of views. Compared to currently stagnating trends such as cryptocurrencies or areas that have faded into obscurity like metaverse real estate, the attention is overwhelmingly stronger. This is because the desire to build machines that resemble humans is a deeply rooted human aspiration.

Contrary to these investor expectations, however, humanoid robots have not yet reached an attractive production cost to become commercialized. Even Chinese-made humanoid robots remain significantly less expensive than their American or Western counterparts, but they are still too costly to implement mass production systems. Analysts say that, for commercial-scale manufacturing to begin, production costs would need to be cut by more than half compared to current levels.

The demand base supporting the humanoid robot industry also remains very fragile. Even if individual consumers purchase out of curiosity, there are very few practical tasks that robots can actually perform for them. The primary customers for Chinese-made robots are not individual consumers, but rather Western robotics companies and researchers. U.S. robotics startups not only draw development ideas from Chinese robots, but also source vital components such as actuators. Over 70% of the world's robotic actuator components are manufactured in China, making the purchase of Chinese robots one of the cheapest ways to acquire actuators, which account for nearly half of a robot’s production cost. That is why American robotics startups visit China to buy robots and components in bulk. In essence, the demand foundation for China's robotics industry is being sustained by the American robotics sector—a market structure that casts doubt on the industry's long-term viability.

The challenge of finding industrial sectors that could create consumer demand for robots is not unique to Chinese companies. Even in Japan, where robotics is highly advanced, there are similar concerns. For example, the "Buddharoid" robot, which was launched in Japan last March and promoted as a robot reading Buddhist sutras, was in reality nothing more than a publicity stunt. Robots used in Tokyo’s senior care facilities primarily function as monitoring cameras to keep track of elderly patients, and less than 10% are actually used for caregiving support.

Despite such unstable demand, more than 150 companies are currently competing within China’s domestic robotics market alone. Unitree is the most recognized among Chinese robotics companies, yet it has already been pushed to second place in terms of global robot shipments. AgiBot, founded in 2023 by former Huawei engineers, surpassed Unitree this year. As new robotics companies sprout up like mushrooms, production is rising—but so is the number of fragile, poorly managed firms.

Driving the quantitative growth of these Chinese robotics companies is the massive policy support provided by the Chinese government. The central government hosts numerous expos to support robotics startups, while local governments aggressively purchase display robots at various local exhibitions. Thanks to these government-driven purchases, Chinese robotics firms saw their revenues increase by more than 50% year-over-year in the first half of this year. However, concerns remain that this sales growth will be unsustainable should the robotics fever subside and government support be withdrawn.

U.S. sanctions against China also present a major geopolitical obstacle for Unitree and other Chinese robotics companies. Last month, the United States imposed a total ban on imports of advanced robots from China, including humanoid and quadruped robots. As over 18% of Unitree’s revenue last year came from the U.S. market, the company is expected to be hit hard. With feedback from major American robotics companies and research institutes now cut off, the pace of technological advancement may also slow.

As a result, even in China—where robotics technology is at the cutting edge—genuine commercial deployment of robots is likely to remain a distant prospect. The kind of robots people really need are not those that dance for publicity, serve beer, or fold laundry once for show. For humanoid robots to become truly commercialized, they must demonstrate their ability to perform dirty, dangerous, and tedious tasks repeatedly, safely, and cost-effectively—precisely the sort of work people tend to avoid.

As CEO Wang emphasized, there is no guarantee that such humanoid robots will arrive in even 10 years. Current marketing pitches around humanoid robotics border on hyperbole, and investor expectations are running far ahead of what the technology can realistically deliver.

Catherine Sokovek, Bloomberg Opinion Columnist

This article is a translation by The Asia Business Daily of the Bloomberg column 'Unitree's Humanoids Have a Long March to Usefulness.'

※This column is published as part of a strategic partnership between The Asia Business Daily and Bloomberg.

This content was produced with the assistance of AI translation services.