The robotics company benchmarking against Unitree is going bankrupt.

The humanoid robot company K-Scale Labs, which was founded just one year ago, is going out of business. According to foreign media reports, Ben Bolt, the CEO of K-Scale, said in an email to investors that "the company has limited remaining operating funds, with a cash reserve of approximately $400,000."

K-Scale was one of the projects that emerged from the YC accelerator. It raised approximately $4 million in a seed round of financing in February 2024, with a valuation of $50 million. The lead investor was the Fellows Fund. Nat Friedman and Daniel Gross, who are well-known for their early investments in Figma and Stripe, also invested $250,000 in this startup in April.

Bolt originally thought it would be easy to raise $10 million to $15 million to keep the company running, but all his efforts ended in failure. He had discussed the possibility of 1X, a humanoid robot manufacturer, acquiring K-Scale with Bernt Bonnich, the CEO of 1X. Unfortunately, the other party was only interested in hiring a few team members and had no interest in the company's other assets. He also had a similar conversation with The Bot Co., another robot manufacturer, but still got no results.

On YouTube, Bolt told the whole story of the company's collapse. Compared with when K-Scale was first founded, the ecosystem in the robot field has changed dramatically. At that time, the cheapest robot cost as much as $90,000, but now Chinese robots have brought the price down, and the industry competition has become more intense.

Indeed, relying on China's supply chain advantages, the robot industry ecosystem is maturing. The Unitree G1, which was released in 2024, starts at a price of 99,000 yuan, and its price in the United States is only about $20,000. Not long ago, Songyan Power released the world's first humanoid robot priced under 10,000 yuan, with a price of only 9,998 yuan. In a report on K-Scale by LatePost, the company regarded Unitree as an important source of inspiration. There were photos of Chinese company teams hanging in the office area to remind the team "how fast Chinese hardware engineers iterate."

A number of robot companies have already fallen in the spring of humanoid robots and embodied intelligence. At an event organized by Silicon Valleyer last year, the CTO of K-Scale once said that the company's motto was "Don't be another failed robot company." Unfortunately, this one-year-old humanoid robot company finally had to shut down.

Got a $2 million order but still couldn't raise funds

Before founding K-Scale, Ben Bolt was an AI researcher at Meta and also worked on Tesla's Autopilot team. Xu Rui, the co-founder and COO, has worked on hardware products at Xiaomi, ByteDance, and Tencent.

K-Scale adopted a strategy of open source and hardware commodification. In May this year, K-Scale open-sourced its humanoid robot software suite and started pre-selling humanoid robots. K-Scale developed a bipedal humanoid robot using open-source software, which allows other robot researchers to train AI models for K-Scale's robots.

The team hopes that these models can endow the robots with various abilities, such as walking in a specific way, performing specific dance moves, or other types of actions. By collecting data provided by open-source contributors, K-Scale can accumulate the diverse real-world data that robot researchers dream of, and this data is crucial for improving robot performance. This may help K-Scale catch up with well-known humanoid robot development companies like Tesla.

However, the competition in this field is also fierce. Other AI companies also have open-source models for robots. For example, Physical Intelligence, a startup that received investment from OpenAI, released its flagship robot AI model π0 in February this year, which anyone can download and use. Xu Rui, the co-founder, once said that compared with robots equipped with the π0 model, K-Scale's robots require less professional knowledge for debugging and setting.

Employees from tech giants such as OpenAI, Nvidia, and Amazon once visited this startup to see its robot products. The outside world once speculated that OpenAI, which is considering developing humanoid robots, might become its ideal customer.

Ultimately, neither customers nor investors showed up as expected. K-Scale originally had about 10 employees, but as the funds ran out, several core employees left the company in recent months. Bolt posted a notice on K-Scale's Discord server, saying that the pre-ordered robots could not be delivered, and the relevant deposits would be refunded. A former K-Scale employee revealed that the company had produced 10 prototype robots, and there were more than 100 orders.

The unit price of these robots is about $15,000, and K-Scale has received orders worth more than $2 million. In the US market, the manufacturing cost of humanoid robots usually reaches hundreds of thousands of dollars, so the price of $15,000 is quite low. However, even though it got a large order, this company still failed to win the trust of Silicon Valley investors.

The first batch of failed humanoid robot companies

At the event mentioned at the beginning, the CTO of K-Scale said that the company's goal was to build a robot platform that was cheap enough and suitable for the general public, so that people could develop their own functions on it. "It doesn't need to be as cool as Tesla's Optimus, nor does it need to be as fully functional as Figure or Unitree. As long as it is useful and easy to develop, it's okay. From this perspective, even with only 10 people and a seed round of financing of a few million dollars, instead of hundreds of millions of dollars, this goal can be achieved."

Looking at it today, this idea was still naive. To build robots, strong capital is a necessary condition. Look at the intensive financing PR in the domestic robot and embodied intelligence fields. For any new company, the financing usually starts at hundreds of millions of yuan, not to mention those big companies that are already valued at billions or even tens of billions of yuan.

In fact, both overseas and in China, a number of robot companies have fallen in the spring of humanoid robots and embodied intelligence.

In February 2025, Aldebaran applied for bankruptcy reorganization due to continuous financial pressure and laid off 50% of its employees. A few months later, the Commercial Court of Paris officially announced that it had entered the judicial liquidation process, marking the complete collapse of this robot enterprise.

Founded in Paris in 2005, this company was one of the first in the world to commercialize bipedal humanoid robots with emotional interaction. In 2012, SoftBank acquired more than 80% of its shares for about $100 million and vigorously promoted its Pepper robot. This robot was once used as a guide in SoftBank stores in Japan, French airports, and Bank of China. In 2020, due to the impact of the pandemic and inventory backlog, Pepper production was stopped, and the business began to shrink. After that, SoftBank sold its European robot business to the German United Robotics Group (URG). Later, URG also stopped providing funds to this unprofitable company.

In April, Embodied, an American AI companion robot company, officially shut down. Founded in 2016, the company's flagship children's emotional robot, Moxie, was once described as "the world's first AI robot friend." The reason for its collapse was also the difficulty in commercialization, and investors were no longer willing to pay for a vague story.

In October, news spread that OneStar, a domestic embodied intelligence startup, had disbanded. According to media confirmation, the company was indeed in the process of disbanding, and "the relevant personnel with Geely background have basically withdrawn."

This was an embodied intelligence startup project initiated by Li Xingxing, the son of Li Shufu. Ding Yan, the co-founder and CTO, was a star researcher at the Shanghai AI Lab and had engaged in research on robot perception and control at Ford, Meta, and other institutions. OneStar was founded in May this year and officially announced a new round of investment in September. The investors included BV Baidu Ventures, NewMargin Capital, Galaxy General Capital, and China-Singapore Group. The reason for its collapse is unknown. Some analysts believe that it was mainly due to insufficient funds, as building robots is very expensive.

Even earlier, CloudMinds Technology, which was valued at $3 billion, was reported to have a broken capital chain and had been in a state of "shrinking to survive" since 2024. In March, Huang Xiaoqing, the founder, chairman, and CEO of CloudMinds, responded publicly that the company was still operating normally, but the business had indeed encountered great difficulties. Therefore, the company had reduced its staff, adjusted its strategy to focus on humanoid robots and cloud brains, and was currently actively raising funds, expanding its business, and trying to solve the problem of unpaid salaries for former employees. However, in the past six months, the company has still been in turmoil, and employees have been unable to get their unpaid salaries.

Huang Xiaoqing revealed in interviews with multiple media that one of the reasons why it has been difficult to raise funds since 2024 is that many startups have flocked to the humanoid robot field, giving investors more choices. They can invest in robot companies in the angel or Series A rounds at a lower valuation. High-valued unicorns like CloudMinds, which are in later financing rounds, are rarely asked about.

Fundamentally, it is an old-generation robot company.

In the domestic market, new robot and embodied intelligence companies are still emerging. However, after more than tens of billions of yuan have been invested in the primary market, the threshold for entering the market has obviously increased. "A company focusing on the robot brain needs to raise at least 2 billion yuan. No matter how excellent the team is, if it fails to raise enough funds, the team is likely to break up." This is the baseline told to me by a partner of a financial advisory firm. The elimination round may have already begun, and funds are flowing towards the emerging leading companies. Only those with sufficiently differentiated technical routes or outstanding backgrounds can survive in this market.

This article is from the WeChat official account "Dongsi Shitiao Capital" (ID: DsstCapital). The author is Liu Yanqiu. It is published by 36Kr with authorization.