The robot was meant to walk through substations at night, where people do not like to go. It was shown on video in September, built to work where gas and dust make a spark dangerous. Two months later the company that built it had stopped paying the people who made it.
No arrest, no accusation of fraud. What is useful here is a dated record of how a robotics startup runs out of road.
Key takeaways
- Wujing Future (无境未来), a Nanjing quadruped maker founded in February 2026, told local media in June it had raised 110 million yuan and signed about 1 billion yuan of orders.
- Employees say wages stopped in August, housing fund payments in September and social insurance in October; 30 to 40 staff are owed close to 8 million yuan (≈ US$1.1M).
- On 30 October, listed investor Junchuang Technology disclosed that its fund's target had wage arrears, suspended operations and stalled R&D, triggering a repurchase clause.
- Most of the headline order book was framework volume and reserve intent, which becomes cash only on purchase orders.
- The same month, China's telecoms research institute warned of bubble risk and predicted a two-to-three-year shakeout.
Eight months, four dates
February. The company registers in Nanjing's Jiangning district.
June. It tells local media it has raised 110 million yuan (≈ US$15M) and signed close to 500 million yuan of framework orders with central state-owned enterprises, plus another 500 million yuan of "local market reserve orders".
August. Salaries stop. The next month housing fund contributions stop; in October, social insurance. Staff begin filing for labour arbitration. By late October the total owed is reported at close to 8 million yuan across 30 to 40 people.
October. A listed investor confirms the damage. Junchuang Technology (骏创科技, 920533), a Beijing Stock Exchange-listed auto-parts maker, tells investors that a fund it backed had put 10 million yuan into the Wujing Future holding company, which now showed wage arrears, suspended business and blocked R&D. The repurchase clause had been triggered; the fund's capital was at risk.
The chief executive told reporters he could not reach the controlling shareholder, who did not answer calls.
What a framework order is worth
A framework agreement in Chinese industrial procurement states a maximum quantity and a unit price, and obliges the buyer to nothing until it issues a purchase order. It is a hunting licence, not revenue. A "reserve order" from a local government or industrial park is weaker still: an expression of intent, often written to support a land or subsidy negotiation.
Add them and you get a number that can be announced, quoted and repeated. It cannot pay salaries.
None of this means the orders were fake. They were contingent — on design freezes, customer budgets, acceptance tests that had not happened. A startup that treats contingent volume as a reason to hire, lease and build runs out of cash in exactly the order above.
Money moved through a fund that first invested in an unrelated medical device company, then swapped equity so the robotics business sat underneath. That structure is normal. It also means "110 million yuan raised" and "110 million yuan in the bank" are different claims, and the public record does not show which was true.
The buyer who walked away
The same period produced a second post-mortem, this one from the customer's side.
Hongqi Chain (红旗连锁, 002697), a Sichuan convenience-store operator, announced in March 2025 that it was developing a robot-staffed unmanned store. The shares jumped the next day and were up about 35% within six weeks.
In July 2026, after 16 months, it terminated the project. The announcement was specific: the arm grabbed irregularly shaped goods unreliably, vision drifted under different lighting, and obstacle avoidance and data transmission were unstable. The system had not reached the standard for mass deployment; the company would scale 24-hour remote staffing instead.
Hongqi's research budget for 2025 was 2.87 million yuan (≈ US$404,000). That is not a robotics programme; it is a feasibility study with a press release attached.
Why this keeps happening
Two failures in a month is no coincidence, and not confined to small companies.
According to a count by tech outlet TMT Post, more than 20 Chinese robotics companies went bankrupt, dissolved or shut business lines between February 2025 and September 2026, across at least 11 subsectors. In October the China Academy of Information and Communications Technology warned that bubble risk was accumulating and that the market would clear out weak companies over two to three years. Funding data agrees: deals fell from 60 in September to 33 in October, according to IT Juzi.
The pattern has the same shape each time. A demo attracts capital. Capital demands orders. Orders are announced as frameworks because that is what is available. Frameworks are spent as if they were cash. Then the calendar catches up.
Honest limitations
The Wujing Future account rests on National Business Daily's reporting and on Junchuang's exchange filing. I could not reach the company; its controller did not respond to reporters, and no audited accounts are public. The 8 million yuan figure is employees' own estimate as reported, not a court finding.
Hongqi's termination is documented in its own exchange announcement. Its R&D figure covers all research, not just the robot project, so it is an upper bound rather than a project budget.
The claim that framework orders are typically non-binding rests on standard practice in Chinese industrial procurement, not on the text of these particular contracts, which are not public.
This is a story about two failures among thousands of attempts. It says nothing about whether the technology works — only about how easily the industry's vocabulary turns intent into revenue.
What readers can do now
- If you are evaluating a robotics supplier, ask what share of the order book is framework, and what share has a purchase order behind it. The gap is the risk.
- If you are an investor, treat "orders signed" as pipeline, not revenue, and check whether the counterparty has budget authority.
- If you are running a startup in this sector, convert one framework into a paid, accepted delivery before you hire against the next ten.
