HomeArticle

Another robotics company has plunged into crisis.

融资中国2026-11-16 10:40
Only when the tide goes out do you find who has been swimming naked.

Recently, multiple media reports revealed that the former CTO, COO and head of product of Amio Robotics have all left their positions. The team has shrunk from a peak of around 130 people to more than 70, and a new round of personnel optimization is still underway. This industrial embodied intelligence company, founded by Liu Fang, former head of intelligent driving mass production at Xiaomi Auto, and led by Anker Innovations, was once regarded as the most promising emerging player to stand out in 2026. Its contraction reflects the starkly polarized reality of the robotics industry.

Since the beginning of this year, Longhui Medical, Suzhou Houzhi, Chuangze Intelligence, Dongguan Lesheng, Hongda Xingye Robotics and Yinghe Robotics have successively had their bankruptcy liquidation applications accepted by courts. Wu Jing Dynamics has suspended operations due to unpaid wages, and Yixing Robotics has been dissolved. From surgical robots to warehousing and logistics, from established unicorns to new companies established for only half a year, market exits have occurred one after another. The problem no longer lies with a single enterprise: an industry propped up by financing narratives is being forced to settle its accounts.

The Onset of Crisis

According to reports from Jiemian News, Anker Innovations added 35.5275 million yuan in equity investment to Amio Robotics in 2025. The real signal is that it did not make any additional capital contribution in the first half of 2026. Once industrial capital stops adding investment, the path for the next round of financing is narrowed by half.

Insiders put it bluntly: the commercialization cycle of embodied intelligence is too long, and Amio Robotics has encountered problems with cash flow. The response from one participating investment institution is only one sentence: the company is still operating normally. This statement means very limited information in the venture capital circle. It usually only indicates that the company has not closed down, and information about whether the core employees are still on board, whether the capital is sufficient, and whether the orders can be renewed is not covered in this sentence.

The crisis of Amio Robotics did not start with the personnel departures, but began when its next round of financing "hit a wall".

From SU7 to Amio Robotics

The resume of founder Liu Fang is the most valuable asset of Amio Robotics.

Liu Fang obtained his bachelor's and master's degrees in computer science from Shanghai Jiao Tong University. He joined Google China in 2006 and was one of the earliest batch of engineers there. In 2012, with these outstanding credentials, he joined Xiaomi, worked on the MIUI Life Yellow Pages, and later took charge of the Xiao AI Speaker, translation machines and in-vehicle rearview mirrors. After Xiaomi Auto was launched, he switched to the autonomous driving business, served as head of product technology and mass production, built the team from scratch, and promoted the whole process to mass production, with the final results applied to the SU7 model.

In the second half of 2024, Liu Fang joined Anker Innovations to take charge of the newly established embodied robotics team, reporting directly to Zhu Fanghao, Vice President of Anker. A few months later, Liu Fang left the company and founded Amio Robotics in September of the same year. Anker later became the lead investor, acting as both the incubation party and the capital contributor.

In March 2025, Amio Robotics, which had been established for only half a year, completed a seed round of financing of about 100 million yuan, led by Anker Innovations and followed by Z Fund (the Zhipu Ecology Fund) and K2VC. Soon after, Sunwoda invested tens of millions of yuan additionally, while CICC Capital and Junshan Capital participated in the tens of millions of yuan angel round. The total financing of the three rounds was close to 200 million yuan.

The financing pace was very fast, but the equity was not diluted much. Liu Fang directly holds 50.5519% of the shares, and indirectly holds 14.8285% through the shareholding platform Amiga, with a total shareholding of about 65.38%. Qichacha shows that Amiga was established in March 2025, and the executing partner is Liu Fang himself. On the institutional side, Anker Innovations holds 8.4253%, Xinglian Dinsen holds 7.9326%, CICC Guanbo holds 5.1709%, Qianhai Hongsheng holds 4.2127%, K2VC holds 1.6851%, and the two Huhan funds hold a total of 5.17%, none of which exceeds 10%.

For an early-stage company that has just completed three rounds of financing, this shareholding structure is not common. Financial investors usually take a higher proportion in the seed and angel stages, but here most of the equity space is reserved for the founder.

It is worth noting that according to the annual industrial and commercial report, the number of people participating in social insurance at Amio Robotics in 2025 was only 23.

There Are Orders, But No Verified Operational Data

Amio Robotics has no shortage of order stories. According to its official website, it completed the first batch of robot deployment in factories in October 2025, and signed its first commercial order and first overseas commercial order in December 2025. Among its customer list, Sunwoda is the most noteworthy. It indirectly holds shares through Qianhai Hongsheng, acting as both a shareholder and a customer. This dual identity of industrial investor and customer is very common in the embodied intelligence industry. The advantage is that the application scenario is ready, while the disadvantage is that the orders are easily regarded as related-party support by the outside world.

The overseas customer is CAMT Automotive from Canada, and the other partner, Diffull Energy, deployed Amio Robotics' solutions to factories in Southeast Asia and the Americas. Diffull mentioned a detail: in the past, production line transformation required a two-week shutdown, but after using Amio Robotics' solution, the product changeover time was reduced to 3 hours. This detail is more illustrative than the order amount, which at least proves that the robots have actually run on the production lines.

In November 2025, Liu Fang disclosed to the media that the company had three KA customers, and the robots were running alongside workers on the customers' production lines. The unit price of a single robot is about 200,000 yuan, which can replace 2 to 2.5 workers, with a payback period of 1 to 2 years. However, none of the information including order amount, number of delivered units, acceptance status, repurchase rate and revenue scale has been disclosed.

At the Shanghai Industry Fair in October 2026, Inovance Technology exhibited the flexible sheet metal assembly solution on the automotive welding demonstration line. The ontology, drive, force control and production line integration are provided by Inovance, while Amio Robotics is responsible for the perception and decision-making layer. Inovance listed Amio as an "AI partner" in its official statement, and thanked it for its support for the smart car welding island demonstration site. This belongs to technical collaboration at the demonstration site level, and neither party has disclosed the purchase amount, number of delivered units and deployed production lines.

Based on public information, the order status of Amio Robotics can be roughly clarified: the customers are real, and Sunwoda, CAMT from Canada and Diffull can all be verified. The scenarios are also real, sheet metal assembly and 3C loading and unloading are actual stations on production lines; there are two versions of the deployment time in factories, August and October, and the difference is not large; the demonstration line of Inovance is a technical collaboration that can prove the company's technical capability. What is missing is the transaction-level data: the amount of each single order, the number of delivered units, and whether any customer has placed a second order, none of which has been made public so far.

It is not uncommon for early-stage hardware companies to disclose information in this way. Customers do not want to make their information public, and the company does not want peers to compare prices. However, now that Amio Robotics has reached the stage of contraction, the same lack of information has changed its nature. What investors care about next is not how long the customer list is, but whether this list can be converted into payment collection within one year.

Stratification of the Industry

Amio Robotics is not an isolated case, it is just a scale on the thermometer.

According to statistics from IT Juzi, in the first half of 2026, 322 financing events took place in China's embodied intelligence sector, with a total amount of about 934.7 billion yuan, exceeding the total amount of 450 billion yuan in 2025, and about 20 new unicorns with a valuation of 100 billion yuan emerged in half a year. IDC data shows that nearly 25,000 humanoid robots were shipped globally in the first half of 2026, of which more than 19,000 were from China, accounting for 77.9% of the global total.

On August 19, Unitree Technology was listed on the Shanghai Sci-Tech Innovation Board, with an issuance market value of about 610 billion yuan, and its market value once stood at 3.2 trillion yuan in November. Megvii was listed on the Hong Kong Stock Exchange on September 1, and Benmo, Xinghuan, Huanchuang and Youdi went public one after another. More than 10 robotics enterprises are sprinting for IPOs throughout the year.

However, IPO is no longer a safety cushion. Since September, 8 out of 14 new stocks listed on the Hong Kong stock market fell below their issue price on the first day of trading, and Maccura fell by 42.93% on the first day. Zhiyuan launched its Hong Kong IPO in July, with a target valuation of HK$40 billion to HK$50 billion, but announced the suspension of the plan in October, and turned to start a new round of financing with a valuation of 50 billion yuan. Even leading enterprises have to wait for the right market window.

On the other hand, the number of failing companies is also increasing. According to media statistics, more than 20 robotics enterprises have closed down in the past 19 months. CloudMinds Robotics has received a total of about 5.4 billion yuan in financing, with a peak valuation of more than 20 billion yuan. Its Beijing subsidiary had its bankruptcy liquidation application accepted by the Fangshan District Court in October 2025. DeepBlue Robotics' application was accepted by the Pudong Court in May 2026, and the creditor's rights of 104 employees exceeded 23 million yuan. The parent company DeepBlue Technology once had a valuation of 14 billion yuan. Yinghe Robotics had its bankruptcy liquidation application accepted by the Shenzhen Intermediate People's Court this summer, and it was once invested by ByteDance and Meituan.

The problem of order caliber is more noteworthy than the number of orders. What is counted as large orders in the industry often includes framework agreements, intended purchases and channel stocking. The signed amount can be very large, but the acceptance, payment and repurchase follow a different set of processes, separated by the actual delivery capability. UBTECH's publicly verifiable contract amount this year is about 1.27 billion yuan, but the delivery guidance for the U1 series has been lowered from the guaranteed 10,000 units to 1,500 to 2,000 units. The contract has been signed, but whether it can be delivered as agreed is another matter. This gap can better reflect the real pace of the industry than the contract amount itself. Wu Jing Dynamics pushed this gap to the extreme: it was established for 8 months, claimed to have obtained 1 billion yuan in orders and 110 million yuan in financing, but the actual delivery volume was zero. It suspended operations due to unpaid wages in October, and its actual controller lost contact.

What Remains of Amio Robotics

Back to Amio Robotics, its technical direction is not wrong. Flexible assembly is the most difficult and most valuable part in industrial scenarios. The combination of first-person video and real-machine reinforcement learning conforms to the current mainstream idea of VLA. Liu Fang's mass production experience is also real, not just resume on PPT.

The problem is that these advantages have not been converted into verifiable commercial data. The company says its technical system is modular, and product iteration will not be interrupted by personnel changes. However, the simultaneous departure of CTO, COO and head of product took away the experience accumulated through running-in with customers.

The confirmed investment loss of 1.539 million yuan by Anker Innovations is not large in absolute terms. According to its 8.4253% shareholding ratio in Amio Robotics, a rough reverse calculation shows that if only the net loss shared under the equity method is counted, this loss corresponds to a loss of about 18 million yuan of the invested enterprise in the first half of 2026. For a hardware company that mainly invests its cash in R&D and delivery, this loss speed is not out of control. Losing 18 million yuan in half a year is not fatal in itself. What is fatal is that the loss is continuing, while the refinancing channel is narrowing.

The one that stopped contributing is the lead investor, not an ordinary financial investor. Anker Innovations not only participated in the incubation, but is also the largest external shareholder, so it knows best about the progress of the company. Its decision not to contribute additional capital has closed half of the door for the next round of financing.

There are only a few remaining paths: seeking new industrial investors or state-owned capital for capital injection, and being acquired is also a possible option. The other choice is to continue to shrink, reduce personnel and expenses to the level that can be covered by existing orders, and survive first.

It still has several cards to play. Sunwoda holds shares through Qianhai Hongsheng and is also a customer, which at least guarantees a stable verification scenario. Diffull has deployed its solutions to factories in Southeast Asia and the Americas, and the demand for production line changeover in overseas production capacity is real. The unit price of 200,000 yuan per robot, compared with the comprehensive annual labor cost of 200,000 to 300,000 yuan for one station in the manufacturing industry, can replace 2 to 2.5 workers, so customers can recover the cost within 1 to 2 years.

What really determines whether these cards can be played out is data. Liu Fang gave a timetable in an interview in November 2025: the company planned to collect thousands of hours of data in that year, and achieve ten thousand hours of data by the end of 2026. According to his own judgment, this magnitude is only the basis for training robots to complete rigid assembly. To be competent in complex work such as flexible assembly, tens of thousands of hours of data are still needed. Even if the node at the end of 2026 is reached as scheduled, it is only the completion of the first stage of the journey.

Another goal Liu Fang set at the same time was to reduce the deployment cycle of new stations from several months to less than one week. These two things are actually two sides of the same coin. Only when enough data is accumulated can the product changeover be completed by on-site workers themselves, without relying on the algorithm team to stay on site. If the goal is fulfilled, the business model will be much more solid than it is now. But this requires time, and what a cash-strapped company lacks the most is exactly time.

This article is from the WeChat Official Account "Rongzhong Finance" (ID: thecapital), author: Dai Xianchao, editor: Wu Ren, and republished by 36Kr with authorization.