If we had to pinpoint the hottest trend in 2024, humanoid robots would undeniably be on the list.

On December 26, GAC Group officially unveiled its independently developed third-generation embodied intelligent humanoid robot, GoMate.

With continuous entries and iterations by various players, the humanoid robot market is becoming increasingly crowded.

In March 2023, Samsung invested over RMB 300 million in local robot manufacturer Rainbow Robotics, whose robot, HUBO, is South Korea's first bipedal walking robot.

From Boston Dynamics, Honda, Tesla, to OpenAI and Microsoft, the advent of the generative AI era has sparked a new wave of competition in robotics.

Humanoid robots are not a novel concept. In 1973, Kazuo Kaneko from Waseda University in Japan led a team to develop the world's first life-sized humanoid robot, WABOT-1. However, half a century later, commercialization remains elusive, primarily hindered by performance and cost challenges.

Robots are comprised of three key technical modules: motion, sensor, and artificial intelligence. For traditional robots, often only one of these technologies is essential. For example, industrial robots focus on motion control, while cleaning robots emphasize navigation and sensing.

In contrast, humanoid robots must be versatile across various application scenarios rather than being task-specific. This complexity necessitates higher technological integration, larger datasets, and stronger language and instruction understanding capabilities. Previously, AI data and models developed in isolation, leading to slow iteration speeds and persistently high costs.

The rise of large models has significantly altered this landscape.

From Transformer in 2017 to GPT-1, BERT, GPT-2, GPT-3, and GPT-4, the number of model parameters has soared from billions to trillions in just a few years. Large models are evolving from single-modality (text, speech, vision) to integrated, general AI, enabling the direct integration of speech, vision, decision-making, and control technologies with humanoid robots, thereby comprehensively enhancing their capabilities.

In April 2023, AI company Levatas collaborated with Boston Dynamics to integrate ChatGPT and Google's speech synthesis technology into the Spot robot dog, enabling human interaction.

The rapid evolution of underlying technologies showcases the potential for large-scale commercialization of humanoid robots, prompting major global tech companies to attempt and prepare for this shift. However, based on the current situation, humanoid robots still have a long way to go before truly becoming a household name.

Firstly, currently available humanoid robots have limited capabilities and lack substitutability benefits.

For instance, home service humanoid robots cannot fully replace human daily services, and in the commercial tour guiding and reception fields, they can only answer simple questions, failing to address all customer inquiries.

Consequently, humanoid robots lacking rigid substitutability are not very appealing to end-consumers.

Participants' product positioning also reveals a current focus on exploring B2B scenarios. For example, Tesla's first batch of mass-produced bots are likely to be deployed in superfactories, while UBTECH focuses on exploring humanoid robot applications in industrial scenarios related to new energy vehicles and 3C electronics in collaboration with enterprises.

Secondly, many fundamental technological shortcomings remain unresolved, such as the need for robot batteries to support up to 20 hours of continuous operation, but most current humanoid robots run for less than 2 hours continuously. Additionally, the high cost, often tens of thousands of dollars, hinders widespread adoption. Manufacturing costs need to be reduced by 15%-20% annually moving forward.

While humanoid robots may not be a profitable venture in the next three years, they are an unavoidable industry in the next 30 years. From an industrial standpoint, the value of humanoid robots lies in replacing high-cost human labor, a highly certain outcome.

Current technical and cost issues will eventually be resolved.

Taking cost as an example, as new technologies or products continue to commercialize, their prices inevitably decline, as seen with computers, smartphones, and electric vehicles. In the past, individual humanoid robots like Honda's ASIMO and Boston Dynamics' Atlas cost as much as $3 million and $1.9 million, respectively. Now, Tesla can achieve a cost of $20,000, with future prices expected to drop further.

Technology is constantly iterating and upgrading. Due to immature lithium battery technology, when Boston Dynamics launched its first-generation humanoid robot Atlas in 2013, it still required a wired power supply. However, by 2016, the second-generation Atlas used independent lithium batteries.

Musk predicts a long-term demand for 10 billion humanoid robots. Even if only one-tenth of this prediction materializes, the industrial space will be enormous.

With the largest population and manufacturing industry, China is already the world's largest robot consumer market, with robust demand from both B2B and B2C users. Moreover, China has not lagged significantly in robot software and hardware technologies. These factors fundamentally determine China's potential to nurture numerous world-class robot companies.

Currently, leading domestic humanoid robot enterprises include UBTECH and CloudMinds, founded in 2012 and 2015, respectively. Realistically, compared to UBTECH and CloudMinds, future prominent domestic humanoid robot enterprises are likely to be major companies like Xiaomi and ByteDance, which have advantages in talent, capital, market, and brand. Although UBTECH started earlier, it hasn't accumulated a strong leading edge, with its humanoid robot revenue currently accounting for only a single-digit percentage.

The upstream sector offers stronger visibility and profitability. Analyzing robot material costs, reducers, servos, and controllers account for 35%, 20%, and 15% of industrial robot costs, respectively, totaling 70%. Considering humanoid robots have more joints and degrees of freedom, these components' proportion may be even higher.

The reducer field boasts multiple players, including Leadshine, Leadshine Harmonic Drive, Tongchuan Technology, Zhongda Lide, and Guomao, all with certain production capacities. However, Leadshine Harmonic Drive stands out as the leader, having established a virtuous development cycle of "research and development - expansion - profitability - further research and development - further expansion." As of 2022, Leadshine Harmonic Drive's annual production capacity reached 400,000 units, expected to expand to 590,000 units in 2023. In comparison, Tongchuan Technology, Leadshine, and Guomao's annual production capacities in 2021 were only 100,000, 60,000, and 30,000 units, respectively. As a manufacturing enterprise, Leadshine Harmonic Drive has achieved a net profit margin of over 30%, a remarkable feat.

The servo motor field exhibits clear class differentiation, with high-end production capacity largely held by foreign companies like Mitsubishi, Yaskawa, Fanuc, and Siemens. Invesense, Jiangte Electric, Jiangsu Leili, Leadshine, and Holzer dominate the mid-to-low-end sector, with Invesense leading the pack. In 2022, Invesense's domestic servo market share reached 21.5%, up 5 percentage points from 2021. With domestic substitution and increased robot production, Invesense's potential expectations are high.

In the controller sector, domestic enterprises are fragmented. While there are professional controller companies like Canopen, Wasion Automation, Googol Tech, Inovance, and Hitech, effective market competitiveness has yet to form, with the current domestic production rate below 20%. It remains to be seen if they will emerge in the future.

In the long run, humanoid robots represent an inevitable future trend. Short-term speculation is futile; what's needed is long-term tracking and attention to identify key enterprises.

Disclaimer

The content related to listed companies in this article is based on the author's personal analysis and judgment, drawn from publicly disclosed information by listed companies in accordance with legal requirements (including but not limited to interim announcements, periodic reports, and official interaction platforms). The information or opinions herein do not constitute investment or other business advice. Market Value Observation shall not be held responsible for any actions taken as a result of adopting this article.

——END——