"Liang Wenfengs" finally get a clear path to IPO.

On June 17, the 2026 Lujiazui Forum opened. Among various agendas, Wu Qing, the chairman of the China Securities Regulatory Commission, dropped a bombshell: the application scope of the fifth set of standards for the Science and Technology Innovation Board (STAR Market) will be extended to the artificial intelligence field.

The fifth set of standards for the STAR Market was originally designed to support technology companies that are not yet profitable but have high-growth potential to go public. After the first round of expansion under the "1+6" policy, it is now being further targeted by regulatory authorities at the track of large AI models.

Now, it is being directed towards a new generation of large model entrepreneurs represented by Liang Wenfeng, opening a new window for a group of AI companies with high R & D investment, not yet profitable, but at the forefront of new-quality productivity.

At the beginning of the year, Zhipu and MiniMax both listed on the Hong Kong Stock Exchange, receiving thousands of times oversubscriptions and seeing their stock prices soar on the first day of listing. Market values in the tens of billions of Hong Kong dollars emerged one after another, boiling up the entire Hong Kong stock market.

A few months later, the two companies separately announced their plans to target the STAR Market. For them, this time they have obtained clearer policy certainty at the institutional level. For companies like Jieyue Xingchen and Yuezhi Anmian that are seeking to go public, it provides more options.

The Scope of the Fifth Set of STAR Market Standards Has Expanded

Wu Qing said in his speech that we should actively embrace the new scientific and technological revolution and industrial transformation, continuously enhance the institutional inclusiveness and adaptability of the capital market, and proposed to expand the scope of application of the fifth set of STAR Market standards to the field of large AI models to support the listing of large AI model companies.

Almost at the same time, the Shanghai Stock Exchange issued the "Guidelines for the Application of the Review Rules for Issuance and Listing No. 10 - Application of the Fifth Set of Listing Standards for the STAR Market by Large AI Model Companies" and solicited public opinions on the Interim Provisions on the Declaration and Recommendation of Issuance and Listing on the STAR Market.

In addition, Wu Qing also mentioned that it is necessary to implement the strategic deployment for the development of future industries, support more "hard technology" companies in fields such as quantum technology, biological manufacturing, and embodied intelligence to list on the STAR Market; relevant guiding opinions on the standardized development of artificial intelligence in the capital market will be issued in due course, and strict investigations and punishments will be carried out on illegal and irregular acts such as hyping up hotspots and concepts under the guise of technology, market manipulation, and insider trading. Illegal stock recommendation, spreading rumors, and illegal trading using artificial intelligence will be severely cracked down on in accordance with the law.

It is worth noting that this expansion of the scope is not just about lowering the threshold. According to the content of the guidelines, the main business of the applicable companies should be the independent R & D, model services, or applications of large AI models, and the core technology products need to belong to the fields encouraged and supported by the national scientific and technological innovation strategy and industrial policies.

That is to say, not all companies with the AI concept can apply. Only those companies with large models as their core technology and whose main business is large models will be supported.

There are also several hard-and-fast rules: at the time of application, there must be at least one large model product that has been launched and achieved large-scale application; while demonstrating their technological leadership, companies must also clarify their advantages in terms of parameter scale, evaluation rankings, and iteration speed; and they must put forward a clear business plan; data security and compliance filing are also essential. Tolerance but not overindulgence is the basic measure of this expansion.

The fifth set of standards for the STAR Market was born at the beginning of the establishment of the STAR Market in 2019. Its biggest feature is to replace the traditional profit requirement with the hard indicators of "market value plus R & D". The main business or products of the company must be approved by relevant national departments, and the market space is large and initial results have been achieved. For companies in the pharmaceutical field, at least one core product must have obtained the approval for the second-phase clinical trial. This is a plan designed for technology companies with long R & D cycles and no short-term benefits.

For a long time, this channel was basically monopolized by innovative pharmaceutical companies. As of the first half of 2025, all 20 companies listed under the fifth set of standards were innovative pharmaceutical companies. Their common profile is: their main products are undergoing clinical trials, they are not profitable, but the clinical data and market scale can support a considerable market value.

This expansion of the channel is a continuous process. In June 2024, the China Securities Regulatory Commission issued the "Eight Measures for the STAR Market" to continue to support the listing of unprofitable technology companies; in March 2025, regulatory authorities clearly and steadily resumed the use of the fifth set of standards; at the Lujiazui Forum held in June of the same year, the "1+6" reform was implemented, adding a science and technology growth layer, restarting the fifth set of standards, and expanding the scope of application from a single field of biomedicine to cutting-edge fields such as artificial intelligence, commercial spaceflight, and low-altitude economy.

Finally, on June 17 this year, the "large AI model industry" was specifically mentioned. This result has been achieved step by step.

Verified Logic and Those in the Queue

One of the reasons why regulators dare to open the door to large AI models is that there has been an exercise that has occurred and been verified by the market.

Back in early 2026, on January 8, Zhipu listed on the Hong Kong stock market, becoming the "world's first large model stock"; just one day later, MiniMax also completed its listing. The enthusiasm for subscription in both IPOs exceeded expectations. The public offering part of Zhipu received thousands of times oversubscriptions, and MiniMax also attracted hundreds of thousands of people to apply for new shares. The financing amounts of the two companies were very considerable.

After listing, the stock prices of the two companies have been rising all the way, and their market values have successively reached the level of tens of billions of Hong Kong dollars. In May and June, Zhipu's market value soared to several hundred billion Hong Kong dollars. The love of the capital market for large AI models was fully demonstrated.

According to the prospectus, Zhipu's total revenue in 2025 was about 724 million, and its R & D expenditure in the same period reached 3.18 billion. The financial statements showed a loss of more than 4.7 billion; MiniMax's revenue in 2025 was about 500 million, and it also suffered a large loss in that year. These two companies are characterized by high R & D investment, large losses, but rapid revenue growth.

The training of the base model and the amortization of computing power are the upfront cost inputs. Measured by the traditional profit threshold, these companies would not be able to enter the market at all; but the "market value + R & D" system of the fifth set of standards can just accommodate them.

The Hong Kong stock market has actually made a pioneering attempt. The examples of Zhipu and MiniMax show that the market is very willing to support a large model company with large losses but a clear technological narrative. In my opinion, this is also an important condition for the STAR Market to follow suit.

Artificial intelligence itself is a business with "front-loaded capital expenditure and post-loaded returns". According to data, the AI-related capital expenditure of several large overseas cloud service providers is still increasing steeply year by year. According to the prediction of Barclays Bank, around 2027, this figure may approach 919 billion US dollars. Under this industry rhythm, if the profit threshold in the domestic market remains unchanged, it means blocking a group of the most strategically significant companies outside.

The introduction of the "Eight Measures for the STAR Market", the "1+6" reform, and the tightening of this policy all follow the same logic: invest in early-stage, small-scale, and hard technology companies, cultivate patient capital, and smooth the cycle of "raising, investing, managing, and exiting". Wu Qing also mentioned in his speech that since the release of the new "Nine National Measures" more than two years ago, the net purchase of A-shares by medium - and long - term funds such as social security and insurance has been considerable. The market value of the A-share technology sector accounts for more than 30%. At the same time, among the companies with a market value of tens of billions, the number of technology companies accounts for 45%.

Currently, Zhipu and MiniMax have successfully listed on the Hong Kong stock market, completed the realization of valuation and capital repatriation ahead of others, and entered the next stage of competition. For their peers that have not yet listed, if they cannot obtain the same amount of funds, they will fall behind in the arms race of "recruiting talent" and "burning computing power". The opening of the domestic channel is not only an effort to keep these companies in the A-share market but also provides an exit option in the domestic market for the primary market.

It is worth noting that on June 1, Zhipu announced that its board of directors had approved the application for listing on the STAR Market; at the same time, MiniMax also issued a similar announcement. The board of directors resolved to explore the issuance of RMB shares, promote the listing on the STAR Market, and signed a tutoring agreement with CITIC Securities. For the two companies that have already listed on the Hong Kong stock market, the expansion of the scope of the fifth set of standards is highly compatible with their idea of "A+H" dual - listing.

Behind them, there is an IPO army seeking to go public. After Yin Qi became the chairman of Jieyue Xingchen, the pace of its capitalization has significantly accelerated. It has successively completed the shareholding system reform and removed the red - chip structure. The market generally believes that it will submit a listing application to the Hong Kong Stock Exchange this year and bet on the direction of "AI + terminal". Yuezhi Anmian has changed its attitude from "not in a hurry to list in the short term" at the beginning of the year to conducting a substantial evaluation of an IPO, and its valuation has also increased accordingly.

If the domestic channel is opened, the two tracks of "Hong Kong stock market + STAR Market" will become clearer, and companies will have more options. This generation of AI entrepreneurs represented by Liang Wenfeng will also have a better chance to ring the listing bell.

The Real Test Has Just Begun

Shifting our focus from the simple "policy benefits", there are more things to think about in this expansion.

In the past, model companies mainly relied on "technological stories" for financing. However, to enter the public market, they need to change their narrative logic from "telling stories" to "realizing commercial value".

During the window period when neither OpenAI nor Anthropic has listed, allowing pure large model companies to be publicly priced in the domestic capital market has significance beyond individual cases. Whoever first establishes a referable pricing anchor for this type of AGI assets will hold a certain amount of say. The performance of Zhipu and MiniMax on the Hong Kong stock market has provided a valuation reference for later - comers; the participation of the STAR Market adds another fulcrum to this "anchor".

The previous generation of companies represented by the "Four AI Dragons" did not end well in the secondary market. Some of them have been under long - term pressure on their stock prices after listing, some broke their issue prices on the first day of listing, some terminated their registration, and some withdrew their IPO applications. Yin Qi, who now heads Jieyue Xingchen, is someone who has come out of that experience, so he repeatedly emphasizes that business must be closed - loop today.

Whether the fifth set of standards can help this generation avoid repeating the same mistakes may not depend on how much money they raise, but on whether the companies can answer the question "where will the future revenue come from".

This is where the risk lies. The guidelines shift the focus of the review from "how much is the current revenue" to "how will the future revenue be generated". Companies need to prove that their target market is clear, the demand is real or has potential, and whether they have advantages in terms of R & D progress and key indicators.

This change leaves room for unprofitable companies, but it also puts forward higher requirements for regulatory authorities and intermediary institutions.

Regulators can accept a company being in a high - investment R & D period and not yet profitable, but they cannot accept it using a vague commercial narrative to replace a verifiable business prospect. A new lesson for sponsoring institutions is to draw a line between an unfalsifiable technological story and a verifiable business logic, which is also the biggest challenge for the entire system.

Balancing tolerance and strict review is like walking on a tightrope. On the one hand, the entrance for hard technology should be opened more clearly; on the other hand, the entrance threshold should be appropriately tightened.

Wu Qing said "strictly investigate and punish those who hype up hotspots and concepts". Considering the hard - and - fast rules in the guidelines, the direction is the same: this expansion of the scope is to support companies with high - tech content and market prospects based on large models, not to open another door for concept hype.

More unobstructed exit channels and fairer international pricing will encourage funds to invest in truly hardcore technologies in the early stage when they are not widely recognized. This is also the feeling of many investors during this period.

The prices and exits in the secondary market have become predictable, and this information will also be transmitted to the primary market, making investment institutions more actively invest in ambitious technology start - up companies.

On the surface, the expansion of the scope of the fifth set of STAR Market standards is an opening to artificial intelligence companies. Deep down, it is a response to "patient capital". The "exit" part, which has been stuck in the cycle of "raising, investing, managing, and exiting" for a long time, is now starting to loosen.

Going back to what Wu Qing said at the Lujiazui Forum, the door is open, but in the end, only a few people who have both high R & D output density and can clarify their business logic can enter.

For this generation of "Liang Wenfengs", there is a door to listing, which is a real positive news; but behind the door awaits a more rigorous review by the capital market than in the primary market.

This article is from the WeChat official account “Rongzhong Finance” (ID: thecapital), written by Wang Tao and published by 36Kr with authorization.