Qingzhou and Yuanrong Secretly Submit Listing Documents, Compete with Momenta for Hong Kong Stock IPO | Exclusive from 36Kr

Image source: Visual China

Text | Xiao Man

Editors | Li Qin, Yang Xuan

36Kr Auto exclusively learned that QCraft and DeepRoute.ai have secretly submitted listing materials to the Hong Kong Stock Exchange and plan to list on the HKEX within this year.

A person in the investment industry told us that DeepRoute.ai submitted its listing materials by the end of 2025, two months earlier than Momenta.

The Hong Kong stock IPO process from submission to listing goes through key stages such as submission review, hearing, roadshow, and share offering. In a smooth scenario, the entire cycle takes about 6 - 9 months. That is to say, this group of autonomous driving algorithm companies, including QCraft, DeepRoute.ai, and Momenta, are very likely to see an IPO boom in the second half of this year.

36Kr learned from multiple people in the investment industry that Momenta's expected IPO valuation will exceed 100 billion yuan; QCraft's latest valuation is in the range of $1.5 billion - $2 billion, and its expected IPO valuation is yet to be known.

"Everyone wants to go public before FSD enters China," said an industry insider. More bluntly, if they are one step slower, these companies may not even have the chance to be "priced."

Intelligent driving suppliers rush to go public, and the window period waits for no one

After the cold winter in 2019, the madness in 2021, and the mass - production stalemate in 2024, the competitive landscape of the intelligent driving track has become very clear. The players still in the game all have mass - production orders from multiple automakers.

Momenta is deeply tied to domestic and foreign automakers such as SAIC, Mercedes - Benz, BMW, Audi, and Dongfeng Nissan. DeepRoute.ai has Great Wall and Leapmotor as its two core customers, and QCraft's customers have expanded from Li Auto to automakers such as Chery and Geely.

Meanwhile, there are structural changes on the demand side. Intelligent driving is changing from an "optional feature" to a "standard feature." In 2026, urban NOA has been introduced to models priced at 150,000 yuan, and even 100,000 yuan.

This means that for the first time, they can close the entire business logic on the financial statements - technology output, mass - production installation, and revenue recognition. The "future stories" repeatedly told by autonomous driving companies years ago are becoming a verifiable reality. For these three algorithm suppliers, going public when the financial statements are the most robust and the order conversion rate is the highest can fetch the best price for their stories.

However, the good times for the intelligent driving industry won't last long. Cao Xudong once judged that "the competition in automotive assisted driving will end in 2026." Li Xiang, the CEO of Li Auto, also proposed in 2025 that the industry will reach the inflection point of autonomous driving within 3 - 5 years. That is to say, the industry will quickly enter a convergence period.

For intelligent driving solution providers, every subsequent battle is related to scale - data scale, mass - production implementation, and customer - binding ability. The one who can get on the market faster, achieve more mass - production, and cover more cities is more likely to form a positive cycle.

The prerequisite for scale is to have enough funds to invest in R & D and mass - production delivery. Large models are devouring resources like a "black hole," and the cost of computing power investment and data closed - loop is increasing exponentially.

Not all companies have the ability to "burn money" in the long run. It can be clearly seen from the financing structures of Momenta, DeepRoute.ai, and QCraft that the investors are mainly industrial capital and automakers, and the investment in autonomous driving from the primary market has significantly cooled down.

The leading algorithm companies can still maintain their investment by relying on the support from automakers and their historical financing reserves. However, for mid - and small - sized players, their cash - flow reserves are not abundant, and with the fierce price war in the industry, their book funds may not be ideal. "It is estimated that they can only hold on for one to two more years," an industry insider analyzed to 36Kr. Therefore, these companies are more in urgent need of opening up the financing channel in the secondary market through IPO to strengthen R & D investment.

This has been verified among automakers. Even though Leapmotor had a mediocre IPO performance, it still managed to get the "ticket" to the secondary market first. Relying on this springboard, Leapmotor not only broadened its fundraising channels but also significantly enhanced its leverage ability and industry influence through the endorsement of the public market, thus pulling ahead of its competitors.

This sense of urgency is becoming more intense with the change of the capital window. Since Pony.ai listed, the window period for Chinese concept stocks in the US stock market has been almost closed by default. Coupled with the high uncertainty of the macro - environment, the financing path has been significantly narrowed. The previously frequently reported plan of Momenta to sprint for an IPO in the US stock market has now shifted to the Hong Kong stock market.

A more realistic threat comes from the change in capital aesthetics. As the global enthusiasm tilts towards "embodied intelligence" and humanoid robots, the secondary market is becoming more stringent in its review of autonomous driving companies.

The time left for autonomous driving companies to sprint into the capital market is running out.

Convergence of technical paradigms, algorithm suppliers need new narratives

Among these companies sprinting for listing, an emerging "anti - consensus" is: The level of valuation no longer directly corresponds to the strength of technical capabilities.

The valuation gradient among Momenta, DeepRoute.ai, and QCraft seems to correspond to the company's size and market position, but the deeper reason is the result of the superposition of financing paths in different time windows.

Momenta completed multiple rounds of financing at the industry's peak and tied up with leading automakers, quickly establishing a valuation anchor. In contrast, DeepRoute.ai and QCraft advanced their mass - production during a more contractionary period and were more restrained in their financing rhythm.

After the industry entered the "end - to - end" paradigm, this gap began to be re - diluted. The current mainstream path has highly converged: centered around large models, the perception, prediction, and decision - making closed - loop are completed through a unified network. Under this framework, the gap in core capabilities among various algorithms has not reached a "generational gap."

Intelligent driving technology no longer provides enough premium space, and enterprises need to find new pricing anchors.

A more comparable reference is the valuation difference between autonomous driving companies and general large - model companies. The combined market value of companies such as Black Sesame, Pony.ai, and WeRide is still less than that of MiniMax, which has been established for less than 5 years (data as of March 31, 2026).

In the eyes of the capital market, autonomous driving has gradually been classified as a vertical application with "high certainty but limited ceiling," while general large models, physical AI, and embodied intelligence correspond to a broader imagination space.

The shift in capital preference is forcing autonomous driving companies to rewrite their stories.

One path is to "go deeper" into the underlying technology stack.

Momenta is extending from a pure software company to the chip layer, emphasizing the "hardware - software integration" system capabilities. This path is highly similar to Horizon, which participates in chip definition and provides a complete "intelligent driving brain + nervous system."

Another path is to "expand outward" and expand the ability boundary.

Both QCraft and DeepRoute.ai are actively downplaying the label of "autonomous driving company" and instead emphasizing the "physical AI" positioning. In this narrative, intelligent driving is no longer the end - goal but just an entry point for intelligence in the physical world.

Zhou Guang, the CEO of DeepRoute.ai, proposed the concept of "RoadAGI" at the NVIDIA GTC 2025, trying to elevate autonomous driving to one of the paths to AGI. According to informed sources, DeepRoute.ai has also started to layout more extensive application scenarios such as robots.

Yu Qian, the chairman and CEO of QCraft, also believes that "autonomous driving is the best entry point to physical AI." When driverless driving truly becomes a reality, it will evolve into general intelligence for the entire physical world.

In essence, they are doing the same thing: turning a "certain market" into a "sufficiently expandable and imaginative future."

Whether these three autonomous driving companies are still an "automotive supplier" or an "AI company," the capital market will not give a standard answer but will make a choice with prices.