The 2024 Hot Track is Getting Crowded

As we approach the year 2024, few can deny the surging interest in humanoid robotsDecember 26th marked a significant milestone in this arena as GAC Group unveiled its third-generation humanoid robot named GoMateWith an ever-growing number of players entering this race, the competition has become increasingly intense.

The momentum gained by the field is evident as major investments flow inFor instance, in March 2023, Samsung made a substantial investment exceeding 300 million yuan in Rainbow Robotics, a local company in South Korea that developed HUBO, the country's first bipedal walking robotThe robotics sector has witnessed contributions from an astonishing range of entities, including renowned names like Boston Dynamics, Honda, Tesla, OpenAI, and MicrosoftThe emergence of generative AI has catalyzed a new wave of competition in the robotics domain.

Humanoid robots are not a novel conception; the foundation was laid decades ago when Ichiro Kato and his team at Waseda University in Japan created WABOT-1 in 1973, the world's first full-sized humanoid robot

Yet, almost fifty years later, the journey towards commercialization remains fraught with challenges, primarily driven by performance and cost concerns.

A humanoid robot is an intricate assembly of three critical technological components: motion, sensing, and artificial intelligenceWhile traditional robots often focus on a specific task leveraging only one of these technologies—like industrial robots emphasizing motion control or vacuum robots focusing on navigation—they do not necessitate the complexity required of humanoid robotsThese robots must achieve versatility to operate within varied scenarios rather than catering to niche tasks, significantly raising the technical stakesThe demand for advanced integration, extensive data modeling, and sophisticated understanding of language and commands renders the journey complex.

Previously, advancements in AI data and models evolved in relative isolation, presenting a slow trajectory in optimization and cost reduction

• Chery's Top Seller Hit by Cancellation Surge
• Inference: The Core Component of Future Applications
• Debate Over U.S. Inflation Outlook
• Federal Reserve Plans Changes to Bank Stress Tests
• AI Toy Craze: ByteDance and Tesla Join the Race

The advent of large models has radically transformed this landscapeWith the inception of models like Transformer in 2017, and the evolution through GPT-1, BERT, GPT-2, GPT-3, and now GPT-4, we've witnessed an astonishing leap in parameters—from millions to hundreds of billions—within a few short yearsThis progression marks a shift towards general AI, integrating various modalities such as speech, vision, and decision-making, thereby augmenting the capabilities of humanoid robots.

April 2023 saw AI company Levatas collaborating with Boston Dynamics to integrate ChatGPT and Google's speech synthesis technology into their Spot robotic dog, marking a clear advancement in human-robot interaction.

The rapid evolution of foundational technologies has ignited a belief that large-scale commercialization of humanoid robots is on the horizonMajor tech firms worldwide are actively exploring this potential

However, the path to mainstream adoption is still largely obstructedCurrently, the capabilities of available humanoid robots remain limited, alongside their lack of effective replacement value for human labor.

For example, household service humanoid robots are far from fully replacing the existing manual services that humans provide, and robots designed for commercial guidance can only respond to a limited set of queries, failing to address the full spectrum of customer inquiriesAs such, without a compelling reason for consumers to invest, the appeal of humanoid robots among individual customers remains muted.

The positioning of market players reinforces this observation, as most innovations are primarily targeting B2B applications rather than consumer scenariosFor instance, Tesla’s bot is likely to find its initial deployment in super factories, while companies like UBTECH focus on applying humanoid robots in sectors such as new energy vehicles and consumer electronics.

Moreover, fundamental technological gaps persist

For example, the batteries used in humanoid robots must support continuous operation for up to 20 hours, while most currently available humanoid robots only function for less than 2 hoursThe financial barrier also remains daunting, with many models priced at tens of thousands of dollars, making mass adoption unlikely unless manufacturing costs decrease by 15% to 20% annually.

Within three years, humanoid robots may not prove to be a lucrative venture, but in thirty years, they represent an industry poised for explosive growthAnalyzing the fundamental value propositions, humanoid robots could replace high-cost human labor, a transition that seems highly probable in the long run.

Addressing the current technological and cost issues may seem daunting, but historical precedents show the acceleration of cost reduction with increased commercializationThe trajectory of computing devices, smartphones, and electric vehicles demonstrates a pattern where high initial prices give way to affordability

For perspective, Honda’s ASIMO and Boston Dynamics’ Atlas originally cost $3 million and $1.9 million, respectivelyPresently, Tesla has managed to produce humanoid robots at costs around $20,000, with expectations for ongoing declines.

On the technology front, progress is also anticipatedBack in 2013, Boston Dynamics had to supply power to the first-generation Atlas robot via cables due to immaturity in lithium battery technologyBy 2016, the second-generation Atlas leveraged independent lithium batteries showcasing how swiftly technology evolves.

Elon Musk has projected a massive future demand for humanoid robots, estimating a staggering reach of 10 billion units globallyEven achieving one-tenth of that figure signifies a market of astronomical potential.

Given its enormous population and robust manufacturing infrastructure, China stands as the world's largest robot consumption market

The demand—both B2B and B2C—will remain strong in the futureAdditionally, China's hardware and software capabilities in robotics have not seen a significant divide, positioning the nation favorably to nurture globally competitive robotics firms.

Leading domestic humanoid robotics companies include UBTECH and Da Vinci Technology, established in 2012 and 2015, respectivelyMoving forward, prominent industry players like Xiaomi and ByteDance, with their substantial advantages in talent, capital, market access, and brand recognition, are expected to play pivotal roles in this spaceDespite UBTECH’s head start, their revenue from humanoid robots remains a mere single-digit percentage, indicating that early moves do not guarantee long-term dominance.

The upstream visibility and profitability within the industry exhibit their strength, particularly when analyzing the material costs of robots

These comprise components like reducers, servos, and controllers, with their collective proportion accounting for up to 70% of industrial robot expensesGiven that humanoid robots require more joints and degrees of freedom, these components' share may be even greater.

The reducers' market features numerous players, with notable names like Dazhu Transmission and LaiFu Harmony alongside othersHowever, one leader stands out: GSK Harmonic DriveThis company has constructed a robust model covering R&D, capacity expansion, profit generation, and reinvestment, achieving an annual capacity of 400,000 units by the end of 2022, projected to rise to 590,000 units in 2023. In contrast, other manufacturers surpassed 100,000 units in 2021, reflecting GSK's dominance in the reduction gear sector.

The servo motor sector is characterized by stark hierarchies, with high-end capabilities mostly in the hands of foreign giants like Mitsubishi and Yamaha, while domestic firms mostly occupy lower-tier markets