Is the Spring Festival Gala robot out of favor? In 2026, robots should do more work and dance less.

In the 2025 Spring Festival Gala, a yangko dance performed by Unitree robots ignited the frenzy in the capital market. In the 2026 Spring Festival Gala, various companies were vying for this rare performance slot. Throughout this year, to attract attention, one show after another emerged incessantly. The audience seemed to have grown tired of such performances. It was nothing more than robots running, dancing, or engaging in martial arts, and the argument that they were just "big toys" kept surfacing.

Looking back on this year, amidst achievements and uncertainties, unicorn companies in the field continuously explored and made trial - and - error attempts. Humanoid robots completed almost all "visible" actions in public: running, dancing, tumbling, sparring... These actions repeatedly refreshed the playback volume on video platforms, but at the same time, they also reduced the value of humanoid robots in the industrial context.

The real turning point lies in this: when all companies can present performances, performance is no longer a competitive edge.

The capital market has been the first to sense this change. While valuations have skyrocketed, orders, deliveries, and large - scale data have clearly lagged behind. There has been an increasingly noticeable gap between technological progress and commercial realization.

This is why the term "bubble" has begun to appear frequently in industry discussions - not because of technological stagnation, but because the criteria for measuring success have quietly changed.

When Show - off Skills Fail

In the process of reshaping these criteria, the Chinese market is not an isolated case. An annual research report from the front - line of the global robotics industry provides a valuable reference.

On January 26th, when China officially announced the list of robots for the Spring Festival Gala, "The 2026 Robotics Industry Outlook" released by The Robot Report used 38 pages of detailed data and interviews with front - line practitioners to outline the real picture of the industry after the hype. For the Chinese robotics industry, which is caught between capital frenzy and commercialization challenges, this report serves as a rare mirror.

The core judgment of the report can be summarized in one sentence: "In 2026, we don't want to see robots dancing or performing martial arts anymore. Show us what they can do that is truly useful." This straightforward statement from Aaron Prather of ASTM International represents the paradigm shift of the entire industry from technology demonstrations to commercial implementation. In the past two years, the promotional videos of robotics companies were filled with eye - catching performances like backflips, splits, and dances. However, when investors and customers start to ask "who will buy the 1000th robot", the click - through rates of these videos cannot be converted into orders. The report statistics show that although there were 463 rounds of financing in the humanoid robot field in 2025 and more than 150 companies entered the market, the actual shipment volume only increased by 17% year - on - year, while the valuation soared by 300% - this disparity is a typical sign of a bubble warning.

The report specifically mentions a trend: as the number of deployed robots increases, enterprises' tolerance for "show - off skills" rapidly decreases, while the importance of reliability, predictability, and total cost of ownership continues to rise. In other words, when robots evolve from "pilot projects" to "daily tools", the evaluation criteria have undergone fundamental changes.

This change is not limited to overseas markets. The Chinese market is going through almost the same process: at the peak of the hype, demonstrating capabilities was sufficient to support valuations. However, after the number of participants increased significantly, only the capabilities that can be integrated into real - world processes began to be repeatedly verified and magnified.

The real watershed occurred in the third quarter of 2025. Agility Robotics' humanoid robot, Digit, became the industry's first product to pass the on - site certification of the U.S. Occupational Safety and Health Administration (OSHA). The significance of this milestone far exceeds any viral video. In an actual e - commerce warehouse in San Antonio, Digit has accumulated over 100,000 pallet - handling operations. The commercial value of a single robot is quantified by simple and straightforward indicators such as "how many paying customers" and "how much cargo has been moved". Meanwhile, Boston Dynamics' debut of the Atlas humanoid robot at CES 2026 was also practical - during the 15 - minute factory parts sorting demonstration, there were multiple instances of manual remote intervention. This honesty won the respect of the industry because it demonstrated the real technological maturity rather than a perfectly edited performance.

The second key shift revealed by the report is the reconstruction of investment logic. From 2024 to 2025, super - large - scale financing (single over $50 million) in the robotics field accounted for 88% of the total capital but only 8% of the transaction volume. This means that capital is concentrating on a few leading companies that are considered to have "commercially proven capabilities". After Figure AI completed its Series C financing, the total committed capital exceeded $1 billion, and the valuation reached $39 billion. Its confidence comes from the first batch of robots delivered to paying customers in December 2024. The judgment of investor F - Prime Capital is more straightforward: in 2026, there will be 3 - 4 clear "winners" in the general - purpose robot track, and the rest of the companies will face the fate of being acquired or shutting down - not because their technology is not advanced enough, but because they cannot find a balance between "high technological uncertainty" and "limited commercial verification".

For Chinese practitioners, the greatest value of this report is not to describe what is happening in the U.S. market, but to provide a new set of evaluation criteria. While domestic capital is still cheering for "another company raising XX billion in financing", the U.S. industry is gradually shifting its focus to three key indicators: payback period (manufacturing requires it to be within 16 months), task - switching ability (whether the robot can expand from single - task to multi - task), and operational data (how much cargo has been actually moved, how many surgeries have been completed). Amazon deploying its 1 millionth robot, Waymo expanding from 4 cities to 5 and planning to enter 15 new markets, Symbolic acquiring Walmart's advanced systems and robotics business for $200 million - the common feature of these cases is that they replace "concept verification" with "large - scale deployment" and "demo demonstration effect" with "customer renewal rate".

More notably, none of the six predictions for 2026 in the report are about technological breakthroughs. Instead, they all focus on business models, supply - chain restructuring, financing strategies, and industry consolidation. This precisely confirms a harsh reality: the competition in the robotics industry has shifted from "whose technology is cooler" to "who can achieve a commercial closed - loop faster". When American peers start discussing leasing models, Robotics as a Service (RaaS), and how to convince customers to accept "not profitable on the first day but with a clear roadmap for task expansion", Chinese enterprises need to think about whether we are ready to answer the same questions from customers.

From the Upper Limit of Capabilities to the Lower Limit of Capabilities

This report does not attempt to make grand technological predictions. Instead, it repeatedly discusses a seemingly simple but life - determining question: how are robots actually used in real - world production and service scenarios?

The report surveyed not only startups but also corporate customers who have paid for robots and are trying to integrate them into their daily operations. In the eyes of these users, there is a clear and cruel dividing line between technology demonstrations and commercial value.

In the past few years, the industry was accustomed to defining robots by their "upper limit of capabilities" - how fast they can run, how high they can jump, and how similar their movements are to humans. However, in actual deployment, customers are more concerned about another set of questions: can it stably repeat the same task?

Does it require the reconstruction of processes and the environment? When it stops working, makes mistakes, or needs maintenance, can the cost and risk be controlled?

These questions may not sound exciting, but they directly determine whether a robot is regarded as a "display device" or "production material".

Against this background, "what exactly constitutes a useful robot" is no longer a philosophical question but a clear watershed - it determines which companies can enter the delivery stage and which will be forced to stay in an ever - extending verification cycle.

Paying for Problem - Solving

Investment institution F - Prime wrote frankly in its annual report: "The market will not pay for novelty but for solving real - world problems. Agility Robotics' Digit moved over 100,000 pallets in 2025. This number is repeatedly mentioned in the report because each pallet - handling operation corresponds to a complete task cycle - identification, grasping, moving, placing, and returning. This proves not just'single - time success' but'repeatability'."

More importantly, these 100,000 pallets were moved in a real e - commerce warehouse, working side - by - side with human employees in a non - standardized environment. When Agility announced this data, Digit had passed the on - site certification of OSHA - evaluated by a nationally recognized third - party testing laboratory at the actual operation site to confirm that the robot remains safe and controllable when human employees are moving around, the lighting changes, or there are water stains or obstacles on the ground.

This is the essence of the shift in the evaluation system: from the "upper limit of capabilities" to the "lower limit of capabilities". In the past, the industry liked to show what robots could do under optimal conditions. Now, customers are more concerned about whether robots will malfunction under the worst - case scenario. Because in a real production environment, the loss caused by an unexpected shutdown may offset the efficiency gains of a month.

16 Months: The Invisible Watershed

The average requirement for the payback period of robot investment in the manufacturing industry is 16 months. This time window may seem like a financial indicator, but it forces a reconstruction of the entire operational logic.

The report cited the calculation method of a procurement manager from a food processing factory: assuming the robot's purchase cost is $100,000, the deployment and integration cost adds another $30,000 - $50,000 (including on - site debugging, process transformation, and employee training), and the annual maintenance cost is about $10,000. To recover the cost within 16 months, this robot must create a quantifiable value increment of about $10,000 per month - whether it is the replacement of labor costs, the increase in production capacity, or the reduction in the loss rate.

This calculation method forces enterprises to answer before purchasing: what exactly is the robot replacing, or what new value is it creating? Vague descriptions like "great potential in the future" hold no ground under this logic. Customers need a specific task list, clear performance indicators, and verifiable operating time.

A more subtle change lies in the customer's definition of "perfection". The report cited a logistics company executive's words: "The robot doesn't need to achieve a return on investment on the first day. If it can reach the benchmark performance and show a clear roadmap for task switching, then the ROI will gradually improve as it takes on more tasks." This reveals an important signal: customers are starting to accept that robots need to "grow", but the premise is that the "growth path must be clear and credible".

This means that enterprises are buying a "capability - expansion system" rather than just a "single - task executor". For example, if a robot initially used for palletizing can learn to unload goods after 3 - 6 months and then learn quality inspection a few months later, its economic value will continue to increase. However, this "task - switching ability" must be verified before deployment - through demonstrations, prior cases, and the transparency of the technical architecture.

The Transfer of Decision - Making Power

The third shift described in the report is more subtle but far - reaching: within enterprises, the dominant power in robot procurement decisions is shifting from the technology team to the finance and operations departments.

In the past, a flashy technology demonstration was enough to persuade the CTO to promote procurement. However, when automation has changed from a "pilot project" to a "daily tool", CFOs and COOs are starting to demand a complete Total Cost of Ownership (TCO) analysis. In some cases, the deployment and integration costs even exceed the hardware itself - which means that a "cheap robot" may not actually be cheap.

This decision - making logic has given rise to new business models. The report shows that 50% of the surveyed enterprises are starting to prefer leasing, and more than 40% are considering RaaS (Robotics as a Service). The logic is straightforward: instead of making a large one - time capital expenditure and bearing the technological risk, it is better to pay monthly and transfer the risk to the supplier. However, the RaaS model places higher requirements on robotics companies: every shutdown is a loss for the supplier; the premise for customers to pay monthly is "continuous availability"; the pricing model must be based on sufficient operational data.

This reconstruction of the evaluation system is essentially an inevitable result of the industry moving from the "story - telling stage" to the "delivery stage". When there are only a few companies in the market, each can define its own standards. However, when more than 150 companies flood into the same track, customers need a unified yardstick to measure who has truly achieved results and who is just talking the talk.

Five Trend Predictions for 2026: The Year of Shuffle

Major Shuffle in Financing

The financing data in 2025 tells a harsh story: large - scale financing accounted for 88% of the total, reaching a historical high, but the number of early - stage small - scale financing transactions plummeted. In simple terms, capital is only flowing to leading enterprises, and startups still in the PPT stage can't even secure angel - round financing.

F - Prime Capital predicts that the number of IPOs of robotics companies will double in 2026, but this is not a sign of spring. On the contrary, it is the last carnival before "escaping the peak". Companies that have received sufficient financing and accumulated some customer cases are eager to lock in their valuations through listing. They know clearly that the window of opportunity will not last long. Meanwhile, the report bluntly predicts that only 3 - 4 enterprises will remain in the humanoid robot track in the end, and the rest will either be acquired or shut down. SoftBank's acquisition of ABB's robotics division for $5.375 billion has already kicked off the integration drama.

China's Opportunities in the Supply - Chain Game

The most dramatic scenario is unfolding: the United States is trying to bring manufacturing back to its homeland through tariffs but finds that it cannot do without the Chinese supply chain; China has the world's most complete robotics industry chain but faces policy barriers in overseas markets.

The complaint of a German supplier is quite representative: "We made large - scale investments in building factories in the United States five years ago, but now we can't buy the key raw materials we need in the United States at all. Tariffs have skyrocketed the cost, and customers are starting to complain." This exposes a paradox: tariffs can prevent the import of finished products but cannot stop the dependence on the supply chain.

This is precisely an opportunity for Chinese enterprises. RealSense revealed that 60% of its customers are using components from the Chinese supply chain - depth sensors, reducers, servo motors, the "internal organs" of robots. The world's most mature supply chain for these components is in China. Even leading U.S. enterprises like Boston Dynamics cannot completely bypass the Chinese supply chain.

A smarter strategy is emerging: Chinese enterprises are deploying production capacity in Thailand and Mexico, using Chinese core components and local labor to serve the target markets. A collaborative robot enterprise in Shenzhen has already set up an assembly line in Mexico. "Global layout, local delivery" may become the mainstream strategy in 2026.

There is also an underestimated track: special - purpose robots. Only a few giants will remain in the general - purpose humanoid robot track, but robots for photovoltaic installation, battery assembly, and textile automation equipment - these vertical scenarios have scattered demand, high customization requirements, and are price - sensitive, which are exactly the strengths of Chinese enterprises.

"Robotics as a Service" Rewrites the Rules of the Game

As mentioned above, a profound change is taking place: 50% of enterprises prefer the leasing model, and more than 40% are willing to try "Robotics as a Service" - paying monthly, including hardware, software, maintenance, and training.

This is not just a simple innovation in the business model but a fundamental shift in investment logic. The manufacturing industry requires a 16 - month payback period, and some enterprises even require 12 months. Buying equipment outright at one time places too much pressure, and there is also the risk of depreciation due to technological iteration. On the contrary, monthly leasing does not occupy the capital budget and allows for upgrades at any time.

The CEO of Apptronik summed it up accurately: "What customers value is not profitability on the first day but task - switching ability. If it can perform Task A today and learn Task B three months later, the way to calculate the return on investment will be completely different." This "growth - oriented investment" logic makes the commercialization of humanoid robots possible.

The "Hidden Demand" Created by the AI Boom

An easily overlooked goldmine: the boom in data - center construction driven by large - language models is boosting the demand for robots. The data - center environment is dense, and the cost of downtime is extremely high. Robots can undertake tasks such as inspection, cable