On July 8th, Momenta, a leading domestic third-party intelligent driving company, was listed on the Hong Kong Stock Exchange. On the day of its listing, the most prominent label in the overwhelming reports was not "autonomous driving," but "the first stock of physical AI."

However, this is not the first time the label "the first stock of physical AI" has appeared in the headlines of Hong Kong stocks.

More than six months ago, on December 30, 2025, 51WORLD was listed on the main board of the Hong Kong Stock Exchange with the stock code "6651," and the banner it raised was "the world's first stock of Physical AI." Six months later, on June 22, 2026, Shenzhen-based multi-spectral AI company HQVT rang the bell for its listing, and external reports changed to "'the first stock of physical AI' HQVT surged 270% on its first day of trading."

The title of "first stock" has been used in turn by three companies within six months. When I took this list to ask a senior person who has been tracking the AI sector for many years, he said two things:

The first: "Non-embodied intelligent large models can no longer gain traction. Capital is collectively looking for an exit, and physical AI is the new story being pushed out."

The second: "But to be honest, none of these three can be considered a proper physical AI company."

He paused and added: "But there’s no way around it; in today’s capital market, the three words 'physical AI' are worth more than 'digital twin,' 'multi-spectral perception,' or 'autonomous driving.' If you don’t rely on it, others will, and their valuation will be higher than yours."

Hat Competition

First, let’s take a look at the "qualifications" of these three companies.

51WORLD was established in 2015 and has long focused on 3D graphics, simulation, and artificial intelligence. It proposed a rather grand vision—the "Earth Cloning Project," which aims to replicate the entire Earth in a virtual world. On its first day of trading, it opened at HKD 35, up 14.75% from the issue price, with a market value that once exceeded HKD 14 billion. The public offering in Hong Kong was oversubscribed by 258 times.

51WORLD is building a full-chain physical AI closed-loop ecosystem of "synthetic data - spatial intelligent models - simulation training platforms." In simple terms, what it does is more like a digital twin, simulating physical laws in a virtual world.

HQVT is headquartered in Shenzhen and focuses on the integration of multi-spectral perception and AI algorithms. What it does is integrate the collection of multi-spectral signals such as ultraviolet, infrared, and visible light, spectral modeling, and intelligent computing to achieve "perception-computation integration."

From a business logic perspective, it is working on the "perception" aspect of physical AI, enabling machines to "see" the physical world. What truly made HQVT stand out is not the technology, but the data. The public offering in Hong Kong was oversubscribed by 7181.21 times, with a total subscription amount exceeding HKD 440 billion, ranking in the top five of the 2026 Hong Kong Stock Exchange main board new stock oversubscription list. On its first day of trading, it surged 270%, with a total market value of HKD 20.19 billion.

Momenta is an intelligent driving company, and what it does is closest to the general understanding of "physical AI," allowing algorithms to understand the real-world physical rules such as the mass, speed, inertia, friction, and motion trajectory of objects around the vehicle Its killer feature is the R7 world model, positioned as the "foundation model of the physical AI era." More than 900,000 mass-produced vehicles equipped with the Momenta system have been launched, with over 210 models accumulated. In the third-party city NOA market, Momenta's market share reached 65%, firmly holding the top position. However, it also has a striking figure: a cumulative net loss of over 9.2 billion yuan in the past three years.

Three companies, three tracks. Digital twin, multispectral perception, intelligent driving, all forcibly crammed under the same label of "the first stock of physical AI."

The senior insider was right; indeed, "each has its own track." So why do they insist on squeezing into the same basket?

Label Inflation

The answer lies in the ambiguity of the concept of "physical AI."

What is physical AI? According to industry consensus, the core of physical AI is the deep integration of physical laws with artificial intelligence, accurately replicating the underlying laws of the physical world in virtual space, achieving dynamic simulation of real scenarios, risk prediction, and decision optimization. Jensen Huang of NVIDIA defines it as the next wave following generative AI. Coatue Management predicts that the market size of physical AI will be at least $6 trillion.

But this definition is so broad that it can encompass almost any AI company related to the "physical world."

51WORLD engages in digital twins, simulating physical laws in the virtual world, which counts as physical AI. HQVT focuses on multispectral perception, allowing machines to perceive spectral information of the physical world, which also counts as physical AI. Momenta works on autonomous driving, enabling AI to understand the motion laws of the physical world, which certainly counts as physical AI.

Anyone can relate to it, and anyone can claim to be "the first."

The term "first stock" in the context of Hong Kong stocks originally defaults to two dimensions: who rings the bell first and who has the most irreplaceable position. The timeline for the former is clear: 51WORLD will be listed in December 2025, HQVT will ring the bell on June 22, 2026, followed closely by Momenta.

However, at the operational level, each company is playing a "modifier game": 51WORLD is the "global first stock of Physical AI"; HQVT is the "first stock of physical AI," omitting "global" while adding an implied modifier of "multispectral AI"; Momenta is the "builder of the physical AI foundation model" + "the first stock of physical AI."

Between the addition and subtraction of modifiers, "first stock" has transformed from a factual judgment into a rhetorical competition.

Even more exaggerated, Panjie Electronics will land on the Hong Kong Stock Exchange on June 17, 2026, with the label of "the first stock of Physical AI edge wireless intelligent chips." The modifiers keep getting longer, and the prefixes keep piling up, turning "first stock" from a scarce item into a fast-moving consumer good.

"The first stock of physical AI" is experiencing a serious inflation However, the first-mover advantage cannot compete with the advantage of being in the right place at the right time. When 51WORLD goes public at the end of 2025, the overall market for physical AI will still be in its infancy, with limited media attention and retail investor interest. Its positioning as the "first physical AI listed company" will only circulate within the primary market and among institutional circles.

Half a year later, the concept of physical AI has been continuously fermented through brokerage research reports and industry forums, leading to a full market explosion. HQVT coincidentally goes public at the right time, combined with the effect of a blockbuster new stock that is oversubscribed by 7,000 times, and intensive market promotion quickly creates a one-sided impression among ordinary retail investors that "HQVT is the first stock." In contrast, 51WORLD, which went public first, finds itself in a passive position in public opinion.

More Valuable than Profit

Ultimately, it is because this label can indeed be converted into money, and it is a substantial amount.

HQVT is the best proof of this.

With 7,181 times retail oversubscription, HKD 440 billion in frozen funds, and a 270% surge on the first day, how much of these numbers is driven by the technical value of "multispectral perception"? Almost zero. Retail investors are voting with their feet, not buying the fundamentals of the company, but the expected premium of the six words "first stock of physical AI." At that moment, the label "physical AI" is more valuable than the company's profitability or technological barriers.

In the past two years, the financing enthusiasm for general large models in China has rapidly cooled, and primary market funds have become cautious. Data shows that in 2025, the number of financing events for large models in China decreased by 23% year-on-year, and the amount decreased by 18%, with leading enterprises accounting for over 80% of financing. Private financing channels have narrowed, making it increasingly difficult for continuously loss-making AI companies to secure funding. Meanwhile, early-stage investment institutions are nearing the end of their fund life and urgently need IPOs to exit. With the primary market becoming stagnant, they must turn to the secondary market for survival.

The Hong Kong Stock Exchange's 18C special technology rules allow unprofitable and revenue-threshold non-compliant hard tech companies to go public, becoming the only way out for these companies. But just because the door is open doesn't mean you can walk out; the question is: going public is easy, but can you sell it?

51WORLD and Momenta have long-term losses, with the latter accumulating losses of over 9.2 billion yuan in the past three years. The stories in traditional tracks can no longer generate premiums, the valuation of digital twins has long been fully priced, and the autonomous driving sector is experiencing a valuation cleanup, with Horizon Robotics and Black Sesame Intelligence having halved from their highs earlier this year. If you say "first stock of digital twins"? No one buys it. If you say "autonomous driving solution provider"? The market is already tired.

But "physical AI" is different.

This is the next wave of AI defined by NVIDIA, a market expectation of $6 trillion, and the most attractive narrative following large models. The same company, when labeled "physical AI," immediately has a different valuation logic.

Thus, Momenta repeatedly emphasizes its positioning as a "builder of physical AI foundational models" in its prospectus and public promotions, aiming to break out of the identity framework of a traditional intelligent driving solution provider and insert itself into a larger track. 51WORLD packages digital twins as a full-chain physical AI closed loop of "synthetic data - spatial intelligent models - simulation training platforms." HQVT packages multispectral perception as the "core of the physical AI perception layer." For investment banks and companies, the "first stock of physical AI" can simplify roadshow narratives, persuade long-term institutions to enter, activate retail enthusiasm for new shares, avoid insufficient fundraising and IPO failures. It raises issuance valuations, attracts cornerstone funds, reduces the risk of issuance failure, and maximizes IPO fundraising scale, ultimately aiming to alleviate cash flow pressure caused by continuous cash burn.

However, the market's awakening came faster than expected. On the third day of listing, HQVT fell over 26%. The premium of the "first stock" could not be sustained. The calm subscription of institutions at 4.81 times and a nearly 30% adjustment in three days point to the same issue: public opinion can elevate your market value to the sky, but the secondary market will ultimately reprice it in its own way, and it may not take long.

Of course, while labels can convert to money, the amount depends on timing. 51WORLD is a counterexample. It went public six months early, and the physical AI concept had not yet ignited public sentiment; even with the "first" halo, it could only circulate within a small institutional circle. In contrast, HQVT went public during the peak of the trend, easily harvesting the national recognition dividend. Being first did not become an advantage; instead, it turned into a "pioneer" that dressed up later entrants.

A senior person said something at the end of the conversation that left a deep impression on me.

He said, "Labels can help you knock on the door of the capital market, but once the door is open, the market looks at performance, profitability, and whether the technology can be implemented. The three words 'physical AI' can help you go public, but they can't help you maintain your market value in the long term."

After saying this, he smiled and added the final sentence: "But first, let's knock on the door."

51WORLD has already been listed, and HQVT and Momenta have also rung the bell one after another. The "first stock of physical AI" label has been worn by all three.

But looking back six months later, who still remembers what the definition of "physical AI" is? Perhaps no one cares. The capital market is never responsible for defining concepts; it only pushes the price of a concept to a sufficiently high level and then looks for the next one.

The game of labels continues. BanJie Electronics has already added modifiers to six words, but it certainly won't be the last. The modifiers will become longer, and the prefixes will stack higher until the three words "first stock" completely lose their premium capability. Only then will the market truly begin to assess the quality of these companies' businesses.

And that window period may come faster than these companies imagine.

The "first stock of physical AI" label is running out, but what is even more scarce than the label is the time these companies have to prove they deserve it. The patience the market gives them may be shorter than expected