At this year's Spring Festival Gala, comedian Cai Ming performed alongside a bionic humanoid robot, while the &#34;Little Bumi&#34; robot moved around backstage, attracting various artists to take photos and exchange New Year greetings.
Both the bionic humanoid robot and the &#34;Little Bumi&#34; robot are products of Beijing Songyan Power Technology Group Co., Ltd. (hereinafter referred to as &#34;Songyan Power&#34;), which currently belongs to two major categories of products—bionic humanoid robots and general biped humanoid robots.
With the brand's continued exposure, Songyan Power's founder Jiang Zheyuan plans to further expand the scale by 2026.
However, challenges still exist.
Automakers Tesla and Li Auto, leveraging their financial and manufacturing advantages, are encroaching on the space of startups, but Jiang Zheyuan remains &#34;cautiously optimistic&#34; about this.
In Jiang Zheyuan's view, the substantial financial resources of automakers are sufficient to support their adherence to &#34;super long-termism,&#34; while the immediate priority for startups is to achieve commercial viability. Due to the misalignment of segmented scenarios and strategic cycles, both sides are currently not competing on the same dimension, and a true head-to-head confrontation may have to wait five to ten years.
The traffic heat that began at the Spring Festival Gala is becoming an opportunity for Songyan Power to accelerate commercialization.
As the competition intensifies, whether Songyan Power can carve out a differentiated path by 2026 is drawing attention.
In a recent dialogue with Wall Street Journal and All Weather Technology, Jiang Zheyuan shared his vision for Songyan Power and the future of the humanoid robot industry.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/2b9824e4-60be-4286-9d45-2d2bb116a3dc.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/2b9824e4-60be-4286-9d45-2d2bb116a3dc.png"/>
<h2>Pursuing a Differentiated Path</h2>
At this year's Spring Festival Gala, Songyan Power's program focused on the more differentiated field of &#34;sketch comedy.&#34;
Comedian Cai Ming performing alongside a bionic Cai Ming made it difficult for the audience to distinguish between the real and the fake Cai Ming for a period of time, avoiding the &#34;red ocean&#34; market of purely physical motion control, and becoming the highlight of the performance.
&#34;Our sketch's biggest punchline is when the 'bionic grandma' appears alongside the real grandma, making everyone realize that the Cai teacher on stage is actually the bionic Cai teacher, which is the biggest twist of the entire program,&#34; Jiang Zheyuan stated.
In Jiang Zheyuan's view, as a hard-tech company rooted in Beijing, having the opportunity to take the stage at the Spring Festival Gala is not just a business opportunity, but also stems from a natural sense of mission.
In fact, this is not the first time Songyan Power has achieved &#34;breaking the circle.&#34;
As early as the 2025 Beijing Yizhuang Half Marathon and Humanoid Robot Half Marathon, the N2 robot developed by Songyan Power won second place, thus &#34;becoming famous overnight.&#34;
However, at that time, facing the sudden influx of traffic and order inquiries brought by the marathon, the early-stage Songyan Power did not have enough energy to handle the &#34;overwhelming wealth,&#34; and &#34;there were regrets,&#34; Jiang Zheyuan admitted.
Having learned from previous lessons, Songyan Power is now well-prepared, with the production team working during the Spring Festival.
Jiang Zheyuan has distilled the strategic core for 2026 into two keywords: &#34;penetration&#34; and &#34;expansion.&#34;
The former refers to the deep cultivation of the existing market. Within the current business landscape, Songyan Power attempts to carve out a larger market share through more refined operations. The latter directly targets those &#34;unexplored areas&#34; that have not yet been fully defined, with K12 education and the consumer market seen as key sources of incremental growth.
Jiang Zheyuan believes that the thousands of well-funded top universities and key high schools are the red ocean that all robotics manufacturers are competing for; beneath the surface, however, are a large number of county-level high schools, vocational colleges, and ordinary universities, which have long been kept out of embodied intelligence due to high price barriers.
Currently, the humanoid robot &#34;Xiao Bumi&#34; under Songyan Power is priced at around 10,000 yuan, and its business logic is designed to match the payment capabilities of this customer segment.
&#34;They may not be able to afford a robot costing tens of thousands of yuan, but a robot priced at 10,000 yuan is something they can buy. Then it can be used to give students a taste of technology, which I believe they can afford,&#34; Jiang Zheyuan stated.
<h2>The Authenticity of the Demand for &#34;Factory Work&#34;</h2>
With Songyan Power, Galaxy General, Yushu Technology, and Magic Atom making appearances at the Spring Festival Gala, the humanoid robot sector in 2026 is undoubtedly set to witness a more intense scale battle.
However, this is not just an &#34;internal war&#34; limited to startups; the battlefield is extending towards automotive companies.
Across the ocean, Tesla has fired the first shot.
In January 2026, Musk announced the discontinuation of the Model S and Model X, the flagship models that established Tesla's status, and will completely transform the factory in Fremont, California, into a production line for Optimus humanoid robots.
Tesla plans to build a production line capable of producing 1 million Optimus robots annually, with production expected to start by the end of 2026.
The cards in Musk's hand are not just about production capacity; in his plans, Optimus will directly reuse Tesla's core AI technology and sensors.
The domestic battlefield is also spreading. Li Xiang, CEO of Li Auto, emphasized in a recent impromptu internal meeting that the company needs to further strengthen its brand positioning in embodied intelligence and will also enter the humanoid robot sector beyond automobiles.
The collective influx of automotive companies into the market to compete with startups stems from a shared anticipation of market explosion.
Currently, the market's imagination for humanoid robots is no longer limited to exhibitions and education but is extending into industrial manufacturing and home services.
UBS Group predicts that global demand for humanoid robots in industrial applications will reach 30,000 units by 2026. Analyst Phyllis Wang believes that as the &#34;brain&#34; of robots evolves, their ability to perform repetitive tasks in workshops has significantly improved. Although currently only a few products can autonomously complete simple tasks, the growth momentum has already been established.
UBS statistics show that among the approximately 18,600 units shipped globally in 2025, Yushu Technology and Zhiyuan Robotics together contributed nearly 10,000 units, occupying a dominant position
However, there is a significant temperature difference between the optimistic research reports in the capital market and the reality in factories.
Although UBS believes that industrial scenarios are the next growth pole, the debate over whether humanoid robots are truly suitable for &#34;working in factories&#34; has never ceased.
When All Weather Technology visited leading domestic textile, cosmetics, and mobile phone manufacturers in 2025, it found that modern factories have a high degree of automation. On a mature production line, robotic arms are responsible for precise operations, and automated equipment can complete multiple production stages independently, allowing one worker to monitor the entire process.
According to an insider from a mobile phone manufacturer in Shenzhen, machines can not only automatically complete the mobile phone manufacturing process but also achieve self-correction and self-verification, improving efficiency while ensuring yield.
Moreover, some mobile phone manufacturers can even develop their own automated equipment based on their understanding of the production line to further enhance production efficiency and reduce costs.
In this highly specialized system that pursues extreme efficiency, the current role of humanoid robots is limited.
&#34;In the short term, we will not invest the most effort into industrial scenarios,&#34; Jiang Zheyuan pointed out.
<h2>Data Remains the Biggest Challenge</h2>
The more imaginative space may still lie in households, where robots can organize clothes, cook meals, and provide emotional companionship like a nanny.
However, household scenarios require robots to have highly generalized capabilities.
An insider from an AI model company in Zhejiang once pointed out to All Weather Technology that households are extremely non-standardized environments. Every type of housing structure, every piece of furniture placement, and even the living habits of each family member vary greatly, posing an exponential challenge to the robot's generalization ability.
&#34;The difficulty of the model essentially comes from the limitations of computing power and algorithms, which is not insurmountable. The real challenge is whether we can collect a sufficiently large and diverse dataset,&#34; Jiang Zheyuan admitted, stating that as long as the data scale breaks through, achieving embodied intelligence is actually &#34;not particularly difficult.&#34;
However, obtaining data in the private space of households is not easy.
Although cleaning devices like robotic vacuum cleaners have begun to collect some environmental data, the high-dimensional data required for humanoid robots (such as visual and operational logic) involves deeper privacy boundaries.
From what All Weather Technology has learned from the industry, some humanoid robot companies use synthetic data for training. For example, some home decoration design AI model companies possess a large amount of physically accurate spatial data, which they use to create virtual home scenarios to provide data training space for robots.
However, whether the quality of this data can meet the needs of robots remains uncertain.
&#34;Especially when it comes to some sensitive data related to family life,&#34; Jiang Zheyuan acknowledged that under the strict compliance with laws and regulations and the protection of user privacy, how to construct an effective channel for data transmission from C-end households to cloud training is currently &#34;not particularly well addressed in the entire industry.&#34;
Perhaps we are still some time away from the grand narrative of humanoid robots completely changing the world, but allowing them to first become accessible tools for some people is an important path to that future
The conclusion of the Spring Festival Gala is precisely the opening for humanoid robots to continue their battles in 2026.
<h2>The following is the dialogue:</h2>
All-weather Technology: Is the humanoid robot an important reason for being selected for the Spring Festival Gala?
Jiang Zheyuan: The humanoid robot is indeed an important reason for appearing on the Spring Festival Gala. The interaction between the bionic Cai Ming and the real Cai Ming in the gala program made the audience aware of the participation of bionic robots in the performance, which is a major highlight of the entire program.
All-weather Technology: What considerations led to participation in the Spring Festival Gala, and was there any internal controversy?
Jiang Zheyuan: Participating in the Spring Festival Gala allows our company's technology, products, and capabilities to be widely disseminated. Through this highly national platform, it helps us break out of our circle and increase the overall brand awareness of the company. There is not much controversy within the company regarding participating in the Spring Festival Gala.
All-weather Technology: How do you make choices between scale and profit?
Jiang Zheyuan: I actually think that for a founder of a company, scale and profit are two things that need to be sacrificed in the early stages.
This year, our primary consideration is still scale, as well as maintaining a good overall cash flow for the company, which is the foundation for us to do all these things.
All-weather Technology: Are there any plans this year that you can share?
Jiang Zheyuan: This year we have two keywords: penetration and exploration.
Penetration refers to increasing product penetration in existing scenarios, which requires adjustments and optimizations in products, solutions, sales teams, and organizational structures. The goal is to become the team with the largest market share in existing scenarios.
At the same time, we will explore untapped areas such as consumer humanoid robots and humanoid robots and solutions for K12 education, aiming to occupy a leading market position from 0 to 1.
In the field of brain technology, our goal this year is to prove the company's capability to build large models from 0 to 1, making a mark on relevant rankings and showcasing the company's strength in AI technology to the market.
In terms of technology research and development, this year we will participate in marathon competitions and invest in more R&amp;D projects to explore more technologies, further consolidating and enhancing the company's position in the field of technology research and development.
All-weather Technology: Why choose the education scenario as a key focus?
Jiang Zheyuan: I think the entire humanoid robot industry is currently in its early development stage. Initially, it presented a situation where there are many scattered scenarios, each not large, but collectively they do add up to a significant amount. So we see this as the most practical and feasible approach at the moment.
We found that if I sell robots to universities and the best high schools in each county, there may only be a few thousand schools.
But if we look below the iceberg, there are relatively less affluent schools that may have many potential customers. They may not be able to afford robots costing tens of thousands of yuan, but a robot priced at 10,000 yuan for technology education for students is something they might be able to afford
All Weather Technology: Previously, you mentioned that you are very frugal with many things, such as requiring your approval for reimbursements over 500 yuan. Is it still the case now?
Jiang Zheyuan: It is still the case. I continue to manage every detail meticulously, and the company maintains a frugal style.
All Weather Technology: As the company grows, are you worried that you won't have enough energy to manage everything?
Jiang Zheyuan: This is something I am very concerned about. Once there are more people, it becomes difficult for me to manage every small detail. However, I am somewhat reassured today because we have brought in some senior executives who can assist me with management tasks.
To be honest, our employees are generally very passionate about this industry, have a high loyalty to the company, and enjoy their current work.
What I hope most is that every employee can work happily. So, considering this, our management difficulty is actually lower, which can also be understood as employees being more self-disciplined.
All Weather Technology: From an industry perspective, is it the time for humanoid robots to scale up?
Jiang Zheyuan: I don't think so. Starting from 2025, it will be the initial exploration phase of scaling up. We can actually see that the delivery scale of humanoid robot companies in 2025 will be around a thousand units.
By 2026, we believe it will be the real beginning of a large-scale explosion, and we expect the entire industry's sales scale to exceed ten thousand units this year.
All Weather Technology: What do you think is the reason humanoid robots cannot yet enter our homes to help with chores?
Jiang Zheyuan: We believe it still depends on breakthroughs in data. If we can collect a sufficiently large and diverse set of data, it is not particularly difficult to achieve.
Currently, the development of intelligent robot brains is not as fast as imagined. What the industry truly lacks is a large amount of high-quality data and new data collection methods. We are preparing relevant solutions to address the issue of insufficient data.
All Weather Technology: Will companies like Li Auto entering the robot space impact the industry?
Jiang Zheyuan: First of all, I still respect Li Auto. Whether car companies will compete with us depends on whether they will touch on some of the scenarios we are currently working on.
What we are doing is more focused on companionship and education scenarios. Whether car companies will really compete with us in such a small-scale industry is unnecessary, in my opinion.
Car companies may be more inclined to believe in some super long-termism. They have sufficient cash flow to support long-term endeavors.
For us, we may need to consider how to achieve commercialization in the short term and realize self-sustainability for the company, so I think there won't be significant competition in the short term.
I believe that if competition arises, it may be in five or ten years

02025

- At the 2026 Spring Festival Gala, comedian Cai Ming performed alongside a bionic robot, showcasing the technology from Songyan Power Technology Group, aimed at increasing brand recognition.  
- Despite fierce competition from established car companies like Tesla, which is shifting to humanoid robot production, founder Jiang Zheyuan remains cautiously optimistic about expanding Songyan's market in 2026.  
- The challenge lies in navigating a crowded industry while focusing on unique market segments like affordable humanoid robots for education and domestic use.

Dialogue with Jiang Zheyuan, founder of Songyan Power: From appearing on the Spring Festival Gala to "seeking scale"