China Is Holding Humanoid IPOs After Unitree Fell 55%

A month ago, Unitree Robotics was the proof that humanoid robots could be a public-market asset. Its shares rose more than fivefold on their Shanghai debut. They have since fallen 55% from that peak, and now Chinese regulators are slowing the companies that planned to list next. According to Reuters, officials have used informal "window guidance" to investment banks to hold back humanoid robotics IPOs. One source told Reuters that listings in the sector were "effectively frozen for now." Another said there is no formal ban, just a slowdown aimed at this one sector.

The line of companies waiting is long. Reuters counts at least six Chinese humanoid firms preparing to go public, including Deep Robotics, X Square Robot and AGIBOT. None of the three answered Reuters' requests for comment. Neither did the China Securities Regulatory Commission.

The question is who is paying

The regulators are not mainly objecting to valuations. They are objecting to where the revenue behind those valuations comes from.

Chinese humanoid companies have earned a large share of their sales from two sources, per Reuters. One is robot data-collection centres, facilities where fleets of robots are operated to generate training data. The other is joint ventures with local governments, which in some cases supply 80% to 90% of the initial investment. When a city government pays for a training centre, and the centre buys robots from the company that co-owns it, the purchase shows up as revenue. Regulators want to know whether any independent customer would buy those robots.

The possible impact is large. A person close to investors told Reuters that valuations at some robot companies could fall 60% to 70% if revenue tied to data-collection centres were removed. The AI Insider, summarising the same reporting, said some private robotics projects have already had valuation cuts of 30% to 50%.

Criticism is also coming from within the industry. Shao Tianlan, CEO of Mech-Mind Robotics, told Reuters that some firms were generating revenue through "data collection centres, related-party deals and other unsustainable arrangements." Mech-Mind has its own reason to care about how the market prices this sector: its shares have fallen nearly 20% from their September 1 debut-day high.

Why Beijing is doing this

Beijing is not turning against robotics. It named "embodied intelligence" a strategic emerging industry, and that designation drew private capital and local-government money into the sector. The IPO slowdown looks like the central government checking whether its own local governments have been inflating demand.

That is our reading of the reporting, and it fits the rest of the market. Mainland Chinese companies have raised $148.9 billion through share sales so far in 2026, up 59% from a year earlier, with technology accounting for 41%, per Reuters. The listing window is wide open overall. Humanoids are the one sector where regulators have narrowed it.

Leo Wang, a venture capitalist at Qianchuang Capital, described the funding wave to Reuters as "campaign-style innovation." He said the hype exceeded what earlier internet and new-energy booms produced, and that some founders were refusing conventional due diligence. A senior banker quoted in the same piece asked: "What's the use case? Is it just people's robots dancing around? Is it working in factories? The volume hasn't really caught up with the hype."

Unitree set the benchmark, then the benchmark dropped

Unitree was the reasonable test case. When its STAR Market listing was approved in July, it was a humanoid maker that was already profitable. It reported fiscal-2025 net profit of 278 million yuan, about $41 million, on revenue of 1.699 billion yuan, and much of that came from quadrupeds and research robots shipped in real volume. If any Chinese robot stock deserved a premium, it was this one.

The stock traded as a symbol of the whole sector. That is why the 55% fall matters beyond Unitree's shareholders. The next several listings were being priced against Unitree, and the companies behind it have less of what made Unitree defensible: a profit record and a customer base independent of government programs. Once the market's reference price fell, every pending offering built on the same story became harder to justify. Regulators appear to have concluded that it is better to stop the queue now than to watch a series of debuts drop after listing.

Ruiying Zhao of S&P Global told Reuters that sentiment had moved from "blanket euphoria to selective rationality." In practice, that means investors will look at deployments, order volumes and products that sell commercially rather than at demo videos.

What it means outside China

For Western humanoid companies, this cuts both ways. Chinese competitors have been able to raise public money cheaply at valuations that no U.S. or European company could match, and that pressure has now eased. The risk is that the reason for the pause, revenue that looks stronger than the demand behind it, applies to the whole industry and not only to China. Pilot programs, paid proofs of concept and strategic-investor purchases are the Western equivalents of a city-funded training centre. They are revenue, but they do not show that a factory would buy a second robot with its own money.

The public-market test for humanoids came faster than expected, and for now it is not going well. Watch three things over the coming months. First, whether AGIBOT or X Square Robot file prospectuses that separate government-linked revenue from other revenue. Second, whether Unitree's share price stabilises above its IPO price. Third, whether the window guidance is lifted quietly or turns into formal disclosure rules. The third would be the more useful result, because it would make the sector report revenue it cannot inflate.