The first-mover advantage vanished overnight. Honor and Xiaomi have entered the market, and the countdown to the survival of second-tier robotics companies has begun.
Why was the highly - regarded Unitree H1 robot outperformed by the Honor Lightning robot in the marathon?
Is it that Unitree robots are not suitable for real - world combat, or is the threshold for humanoid robots so low that even major mobile phone manufacturers can easily create them?
Recently, at the Beijing Yizhuang Humanoid Robot Half - Marathon, the Honor Lightning robots took the top three positions.
Leading companies like Unitree Technology and the top two finishers of the previous event, Tiangu and Songyan Power, did not perform perfectly this year and were overtaken by the “newcomer” Honor.
Ironically, the Honor robot department was established around June 2025. In just about a year, it could outperform the leading players. Second - tier players such as Magic Atom, Accelerated Evolution, Half - Awake Technology, and Youbaote are far behind.
Behind Honor Lightning's victory, a possible change is that the first - mover advantage in the robot industry is rapidly disappearing. With the in - depth layout of “cross - border players”, it may have a devastating impact on the humanoid robot industry. Especially for second - tier brand manufacturers, the situation will be even more difficult.
The first - mover advantage vanishes overnight, and big manufacturers break through the industry's “moat”
Unitree Technology, which shone on the Spring Festival Gala stage, was a favorite to win the half - marathon before the event. However, its performance in the actual competition was disappointing. The main Unitree H1 robot lost control and fell, and was carried off the track. Another G1 robot modified by a third - party also had problems such as veering off course and falling several times.
On the contrary, the Honor Lightning robot, which was not initially favored, successfully counterattacked and took the top three positions. Honor achieved in just over a year what many robot companies had explored for years, and even outperformed many “senior” players.
Does this mean that the threshold for humanoid robots may not be as high as expected? Have the original leading players lost their advantages?
To build a good humanoid robot, the most difficult part is to ensure the whole - body coordinated movement ability, which involves multiple core technologies such as motion control, servo joint control, dexterous hands, the whole - machine system, thermal management, and environmental perception. It requires strong software - hardware integration capabilities.
Honor was able to “take the lead” in just over a year, which is actually a result of technological equalization.
In fact, as the popularity of humanoid robots continues to rise and the supply chain matures rapidly, it is no longer difficult to create a robot.
In addition to independent humanoid robot start - up companies, Internet giants represented by Tencent, automobile manufacturers represented by BYD, GAC, and XPeng, mobile phone manufacturers represented by Honor, Xiaomi, and vivo, industrial robot manufacturers represented by Estun, Aft, Siasun Robot & Automation, Yuejiang Technology, and JAKA Robotics, service robot manufacturers represented by Keenon Robotics and Pudu Robotics, lidar companies represented by RoboSense, and some consumer electronics precision manufacturing manufacturers and artificial intelligence and autonomous driving companies are all eager to enter the market.
In terms of hardware, the localization rate of core components such as precision reducers, high - performance servo motors, and sensors has reached over 80%, breaking the overseas monopoly. The iteration speed has changed dramatically.
Regions such as the Beijing - Tianjin - Hebei region, the Pearl River Delta, and the Yangtze River Delta have formed a complete industrial chain covering core components, whole - machine manufacturing, and application services, continuously reducing manufacturing costs and integration difficulties.
Public information shows that the Honor robot body is built by Lens Technology, equipped with Huake Cold Core high - speed suspension pumps, Orbbec Gemini 330 series binocular 3D cameras, Hesai JT128 lidars, and Lingyi Precision Industry, AAC Technologies, etc. have provided hundreds of core components.
The scary thing is that Honor didn't specifically look for specialized robot suppliers. Many suppliers like Lens Technology, Lingyi Precision Industry, and AAC Technologies are also players in the mobile phone industry chain.
In other words, Honor simply migrated its mobile phone industry chain capabilities to the humanoid robot industry and quickly created the “Lightning” that can outperform many players.
In terms of software, the late - comer advantage can also be seen. In the past, it took a long time to accumulate technologies such as algorithms. But today, with the rapid development of large models and intelligent agents, the training cycle for humanoid robots has been significantly shortened.
With continuous improvement in the “brain” intelligent decision - making system and the “cerebellum” motion control algorithm, combined with a comprehensive breakthrough in the hardware supply chain, currently, the humanoid robot industry has gradually lost its long - standing “moat” and initially has the basic conditions for rapid integration.
Unitree Technology, Zhipu Robotics, etc. all adopt heavy - asset models. In contrast, Honor is very smart and easily achieved a comeback by standing on the shoulders of predecessors.
The survival environment of second - tier body manufacturers is deteriorating rapidly
“We don't need to over - mythologize the humanoid robot marathon itself, but we can't ignore the rapid changes in the industry and enterprises behind the marathon,” an industry expert told EqualOcean.
The head of a market - oriented VC told EqualOcean, “From a longer - term perspective, every industry and track has its own cycle and fluctuations. The humanoid robot industry has been popular for a long time, which is quite rare in China's primary investment market. Currently, hundreds of enterprises have flocked to the industry, which is obviously over - heated. We also think the investment in the primary market is crazy.”
“We previously predicted that the first wave of differentiation and elimination in the humanoid robot industry would occur in the first half of 2025. After the first robot marathon, although there are still many non - consensuses in the industry, judging from the flow of funds, many robot and industry chain enterprises benefited from this event, and the industry differentiation was postponed by more than a year. After Honor won the championship in the second event, many investors became more cautious, and the real test will come in the second half of this year.”
Honor's sudden success caught humanoid robot enterprises off - guard. Leading robot enterprises represented by Unitree and Zhipu, with the support of funds, customer resources, technological accumulation, scenario data, and brand influence, have at least crossed the survival safety line in the short term.
However, the large number of mid - and small - sized manufacturers will have a hard time in the future.
On the one hand, they originally lacked leading technologies. After the first - mover advantage in the industry gradually fades, they will soon be caught up by late - comers.
On the other hand, it will be more difficult for mid - and small - sized enterprises to raise funds. The “2026 China Embodied Intelligence Enterprises Going Global Panoramic Report - Capital Craze and Globalization Inflection Point” released by EqualOcean Intelligence shows that in the first quarter of 2026, a total of 197 financings were completed in the Chinese embodied intelligence track.
The data clearly shows that funds are flowing to in - depth players in sub - fields such as sensors. In the past, the investment was “spread out”, but now it is “concentrated on the top”. And enterprises with core breakthrough capabilities are obviously more favored by capital.
In addition, it is worth mentioning that many humanoid robot enterprises still have no revenue, no mass production, and lack delivery capabilities. The only prototypes they have are just for show, and it is quite difficult to complete actions such as side - kicks, backflips, and long - time running. Such enterprises will find it difficult to attract financing with prototypes and stories.
Specifically for enterprises, the humanoid robot industry is large - scale and has many players, so it is difficult to have a unified indicator or standard for classification.
If we use valuation/market value as an indicator, according to EqualOcean statistics, Chinese embodied intelligence enterprises with a valuation/market value of over 10 billion yuan are: Ubtech Robotics (about 45.9 billion), Unitree Technology (about 42 billion), Galaxy Universal (about 21 billion), Xinghaitu (about 20 billion), Tashizhihang (about 18 billion), Zhipu Robotics (about 15 billion), Yuejiang Robotics (about 11.7 billion), Qianxun Intelligence (about 10 billion), Zhifang (about 10 billion), Independent Variable Robotics (about 10 billion), Yun Shen Chu Technology (about 10 billion), Zhongqing Robotics (about 10 billion), Xingdong Jiyuan (about 10 billion), Keenon Robotics (about 10 billion), Pacini Sensing (about 10 billion), Fourier Intelligence (about 10 billion), Zhujidongli (about 10 billion). These enterprises are relatively leading in terms of products, technology, financing, and commercialization, and can be regarded as first - tier manufacturers. (Note: The data is as of April 29, 2026. The valuations of non - listed enterprises are estimated based on the latest financing and public information.)
But there are more mid - and small - sized manufacturers that have not entered the “10 - billion club” (including those incubated by universities): Lingbao Robotics, Leju Robotics, Songyan Power, Youbaote Robotics, Magic Atom, Luming Robotics, Tianlian Robotics, Jingchu Robotics, Iron Man Technology, Zhuoyide Robotics, Qingbao Robotics, Kepler Intelligence, Qingxinyichuang, Daimeng Robotics, Cyborg Intelligent Future, Qizhi Robotics, Accelerated Evolution, West Lake Robotics, Yuanli Infinite, Puluo Universe, Glano Robotics, Wujie Exploration, Mojia Zhichuang, Bilingxing Technology, Zhijian Power, Wuba Intelligence, Haohai Xingkong, Ruoyu Technology, Digital China, Ruantong Tianqing, Xingsi Wujie, Yunmu Zhizao, Youliqi Robotics, Shuangzi Zhiqing, Xinggong Jujang, Lingdong Universal, Zero Power, Zhongke Guiji, Yuequan Bionics, Half - Awake Technology, etc.
In the industry, mid - and small - sized players can hardly surpass leading brands like Unitree Technology and Zhipu Robotics for the time being; externally, they will also face more cross - border players like Honor, which means they are in a difficult situation.
Songyan Power, which shone at the first humanoid robot marathon in 2025, was a direct beneficiary of the event. Just when everyone thought Songyan Power would continue to improve on this track, it shifted its business focus to the home service scenario.
An industry insider told EqualOcean, “Songyan Power's shift is actually based on a clear judgment of the industry situation. They found that as leading players like Unitree expanded their advantages in the motion control field, this path became too competitive. Later, they even spent a lot of money to appear on the Spring Festival Gala stage and perform a skit with Cai Ming in 'Grandma's Favorite' to strengthen this home service scenario.”
In addition to Songyan Power, Kepler Robotics also shifted from an early - stage robot hardware manufacturer to an “industrial embodied intelligence data service provider”. “Many early - stage body manufacturers, after failing to reach the level and scale of Unitree and Zhipu in the short term, can only quickly turn to new directions to find new breakthroughs and get the support of the capital market. After all, the leading players are preparing for listing. On the one hand, the capital market cannot accommodate so many robot body companies. On the other hand, after the capital market absorbs new concepts, it will ultimately come down to performance. Our investment focus in the robot industry chain has shifted to core components, the 'brain'/'cerebellum', and we also pay a little attention to new materials and new processes related to the robot industry.” an investor of a robot industry fund revealed in an interview with EqualOcean.
EqualOcean believes that in the early stage of the industry, telling stories can attract capital, but after the “tide recedes” and the industry experiences a bubble, it will inevitably return to common sense and rationality.
For most second - and third - tier body manufacturers, if they cannot make rapid breakthroughs in technology, scenarios, brand, delivery volume, supply chain, etc. in the short term and truly achieve large - scale implementation, once the financing window closes, their fate is easy to predict.
Since Honor can “take the lead” in just one year, how much time is left for second - tier robot manufacturers?
There is no other way but to move from performance to real - world combat.
Cross - border humanoid robots, big manufacturers' devastating impact
Honor Lightning's victory has shown the outside world a new change in the humanoid robot industry. If Honor can succeed, why can't major mobile phone manufacturers like Huawei, Xiaomi, OPPO, vivo, and ZTE “follow suit”?
The mobile phone market is highly competitive, and the growth space is constantly shrinking. Mobile phone manufacturers are facing difficulties. Cross - border development is inevitable.
The humanoid robot industry has huge imagination space and relatively high product gross margins, with very considerable profits.
As disclosed in Unitree Technology's prospectus, from 2023 to 2025, Unitree Technology's gross profit margins of its main business were as high as 44.22%, 56.41%, and 60.27% respectively.
More importantly, there are many transferable technologies between mobile phones and humanoid robots, including batteries, SoCs, thermal management, mass - production processes, and supply chains.
Honor's entry into the field shows that major mobile phone manufacturers are fully capable of quickly creating humanoid robots.
For example, Xiaomi entered the robot industry early. In 2021, it launched the quadruped robot “Tiedan” CyberDog. In March this year, Lei Jun said that Xiaomi's robots have started “internships” in automobile factories and have initially achieved autonomous work in scenarios such as self - tapping nut loading stations and material box handling.
On April 27, Xiaomi's new humanoid robot was publicly unveiled at an investor event, continuing the humanoid design of CyberOne in 2022. Given time, Xiaomi may also become the next “dark horse”.
Mobile phone manufacturers are just one example.
In the past, people might think that the competitors of humanoid robot enterprises were other robot enterprises. But now, this perception is being subverted. Maybe the competitors are no longer “another” but “others”, coming from cross - border players such as Internet giants, automobile manufacturers, mobile phone manufacturers, and consumer electronics giants.
Especially in terms of data, scenarios, capital, and supply chains, big manufacturers have natural advantages. Many of their existing businesses are high - frequency scenario entry points. Once they are open - sourced, they can attract external developers to contribute data, and the advantages such as cost - effectiveness and performance parameters that robot start - up enterprises rely on will soon be diluted.
For mid - and small - sized players, what advantages are left after these giants enter the market? What competitiveness do they still have? Can their business models still hold? These are questions that must be faced directly.
A new world, just beginning
The development speed of humanoid robots far exceeds expectations.
At last year's Beijing Yizhuang Humanoid Robot Half - Marathon, many participating robots had to be remotely controlled by humans to run, and the completion rate was less than 30%.
However, just one year later, there are now more than a hundred participating teams and over 300 humanoid robots competing on the track, and nearly 40% of the participating teams use autonomous navigation.
While more brands are daring to show their strength, the running speed of humanoid robots has also increased significantly. Honor Lightning refreshed people's perception with a time of 50 minutes and 26 seconds, which is two - thirds faster than last year's champion, Tiangu Ultra (2 hours, 40 minutes, and 42 seconds).
It's easy to imagine that in 2027, more “dark horses” will emerge on the track. Who will win the championship? It's all unknown.
But one thing is certain: the human - machine era is just beginning, and a new world is approaching.