Why are Zhipu and MiniMax in a hurry to return to the A-share market?
On May 29th, Zhipu's market value soared to HK$800 billion at one point during intraday trading. With its massive scale approaching one trillion, it publicly declared the birth of a new generation of AI giants.
On the same day, MiniMax submitted its filing for A-share listing guidance.
One is racing ahead in the capital market. The other is going back.
On the surface, these are two separate capital trajectories evolving independently.
However, if we dissect their financial fundamentals and timelines, they have both made the same fateful new choice.
I. Why Exactly Now?
Zhipu was listed on the Hong Kong Stock Exchange on January 8th, and MiniMax followed one day later. No large model companies went public on the A-share market during the same period.
It wasn't until May 29th that MiniMax's announcement completely revealed the iceberg beneath the surface.
During this four-month-and-twenty-day vacuum period, Zhipu skyrocketed 13 times under the global liquidity premium in the secondary market, and MiniMax also achieved a 4-fold increase against the odds. These two leading companies not only hold huge amounts of US dollars in the primary market but also raised sufficient funds on the Hong Kong Stock Exchange. They have never been short of money.
So why didn't they directly launch the dual-listing structure of A+H on the day of listing? Why exactly now?
The secret lies in the months-long capital incubation period of the two companies.
In 2023, Chinese AI companies were called "Chinese OpenAI", and no one knew how much they were worth. Investors made decisions based on intuition, and founders accepted prices based on their confidence. The entire primary market was in a state of collective excitement. As long as the team had an impressive resume, the valuation could skyrocket.
In 2024, people found that the money was burning faster than expected. Each round of model training cost hundreds of millions, and the inference cost couldn't be reduced. The pace of financing began to slow down. Some companies that originally claimed to aim for a valuation of tens of billions of US dollars quietly lowered their valuations.
In 2025, the entire industry was forced to enter a year of commercialization exams. However, due to the lack of real samples in the public market, the valuations of dozens or even hundreds of billions of unicorn companies such as Yuezhianmian, Baichuan Intelligence, and Lingyiwanwu were still just numbers determined behind closed doors. No one could prove them wrong, and no one dared to guarantee them.
The entire Chinese large model industry is stuck in an extremely awkward situation: the exciting stories are told, but the fair valuations can't be calculated, and the primary capital is unable to continue supporting.
It wasn't until Zhipu and MiniMax broke through the situation on the Hong Kong Stock Exchange first. The global public capital gave the real pricing of large model assets for the first time.
This is more important than any research report.
It is precisely because of the high-liquidity testing ground of the Hong Kong Stock Exchange as a valuation reference that the regulators and domestic institutions in the A-share market finally saw clearly the pricing model of this new type of asset.
This is the underlying logic for the two giants to choose to go back one after another. The Hong Kong Stock Exchange is responsible for establishing the valuation reference, and the A-share market is responsible for taking over the subsequent assets. The sequence cannot be reversed.
If they had submitted applications for the A-share market on the day of listing on January 8th, it would have been like forcing the pricing issue to the domestic market. Most likely, it would have been stuck in the review process. Facing unprofitable AI companies without comparable benchmarks and mature profit models, the regular review of the Science and Technology Innovation Board could only be postponed indefinitely.
Another policy foreshadowing that cannot be ignored is that in July 2025, the China Securities Regulatory Commission officially issued an opinion, carrying out a landmark targeted expansion and regulatory clarification of the scope of application of the fifth set of listing standards on the Science and Technology Innovation Board, opening up the listing channel for artificial intelligence companies to enter the Science and Technology Innovation Board at the institutional level.
Zhipu was more sensitive than anyone expected. It didn't wait until the Hong Kong Stock Exchange operations were smooth before taking action. As early as April 2025, when the primary market was in a collective frenzy, Zhipu quietly submitted its first version of the A-share listing guidance filing, three months earlier than the bonus policy of the fifth set of standards on the Science and Technology Innovation Board in July.
An even more interesting change occurred in February this year. The IPO guidance progress updated on the official website of the China Securities Regulatory Commission showed that Zhipu, which was already worth hundreds of billions, made an extremely rare move. It withdrew the initial listing guidance filing submitted in April 2025 and immediately handled a new listing guidance registration, aiming directly at the Science and Technology Innovation Board.
This is not a retreat. It is an active adjustment and a new start after seeing clearly the trends of the capital markets on both sides.
There is the policy bonus on the left, the Hong Kong Stock Exchange pricing on the right, and the A-share listing guidance following closely.
Reconstructing these three actions into a timeline, this is not a sudden idea of a certain founding team. It is a precise operation of a capital market relay pipeline, which, on May 29th, 2026, passed the baton to the third leg exactly on time.
II. The Hong Kong Stock Exchange Provides Valuation, and the A-share Market Provides Position
What the Hong Kong Stock Exchange provides for MiniMax and Zhipu is an entry ticket that the world's top long-term capital is willing to publicly pay for. Sovereign wealth funds from the Middle East and Europe, such as the Abu Dhabi Investment Authority, Singapore's GIC, and the Norwegian Central Bank, have focused on investing in MiniMax and deploying in the domestic AI track.
On May 22nd, the Hang Seng Tech Index officially announced that the two companies would be included in its constituent stocks simultaneously. After the relevant rules take effect on June 8th, a large amount of passive index funds will be automatically locked and allocated according to the index rules.
This is the irreplaceable global liquidity advantage of the Hong Kong Stock Exchange as an offshore financial center.
What the A-share market provides is the fundamental for hard-tech companies to survive in this special era: Position.
Behind the position are three layers of intangible assets that cannot be calculated by a simple financial model.
The first layer is the national strategy. The Science and Technology Innovation Board is positioned as the main battlefield for hard technology. Entering the Science and Technology Innovation Board means being officially recognized as a national strategic science and technology asset. This label has actual weight in the Chinese business environment. It will affect the government procurement list, the cooperation willingness of central and state-owned enterprises, and the priority of obtaining computing power indicators at critical moments.
The second layer is long-term funds. Social security funds, insurance funds, national teams, and local state-owned asset funds. These funds won't leave due to short-term fluctuations. For a long-term loss-making AI company, they are truly patient capital. The money in the Hong Kong Stock Exchange is smart but lacks patience. Once the overseas macro situation turns, foreign capital will withdraw faster than anyone else.
The third layer is industrial support. From computing power indicators to data compliance, from government procurement to industry standard setting, the industrial resources that an A-share Science and Technology Innovation Board company can obtain are beyond the reach of a Hong Kong Stock Exchange-listed company. In the end, what matters in the AI field is not just the model ability but whether the entire national industrial chain is willing to run with you.
The Hong Kong Stock Exchange provides real and liquid money. The A-share market provides an irreplaceable strategic position. Money can be raised again. But the position, in this era, only has one chance to seize.
There is also a detail worth highlighting, the choice of sponsors.
For MiniMax's Hong Kong Stock Exchange IPO, China International Capital Corporation (CICC) is one of the joint sponsors, and it signed with CITIC Securities for A-share listing guidance. For Zhipu's Hong Kong Stock Exchange IPO, CICC is the sole sponsor. In the early stage of A-share listing guidance, CICC was the sole provider, and in February 2026, Guotai Junan and Haitong Securities were added for joint guidance. Both leading AI companies are associated with CICC in the Hong Kong Stock Exchange and have selected top domestic securities firms for the A-share market.
This is not a simple choice of investment banks. CITIC Securities is the leading sponsor for the A-share Science and Technology Innovation Board, with a large number of hard-tech IPO projects completed over the years. Guotai Junan and Haitong Securities are industry giants born from the merger of securities firms in 2025, with prominent state-owned asset resources and policy coordination capabilities. One leads the industry in the ability to implement market-oriented projects, and the other has significant advantages in top-level regulatory coordination.
This combination in turn confirms one thing: the path back to the A-share market is far more complex, intense in competition, and strategic in importance than the Hong Kong Stock Exchange.
III. This Is Not an Isolated Phenomenon
In the past decade, the most outstanding Chinese Internet companies have almost followed the same path.
Tencent went to Hong Kong. Alibaba, JD.com, and Pinduoduo flocked to the United States. ByteDance has been waiting for a long time amidst the overseas geopolitical tug-of-war.
The capital dream of that generation of companies was Nasdaq. It was the American English at the roadshow, the conference calls with Wall Street analysts, and being heavily held by overseas long-term funds. The A-share market couldn't accommodate them at that time. Each issue, such as the VIE structure, loss-making listing, and different voting rights for the same shares, was an insurmountable obstacle. They were forced to go overseas.
Their move to the United States back then was a passive exile under the old system. Today, this group of top AI companies is taking the opposite path.
First, they get a global valuation on the Hong Kong Stock Exchange, and then they get a national position on the A-share market. Nasdaq, which once symbolized the totem of technology, has been completely removed from their strategic table. Even if it is still on the table, its weight has decreased significantly.
This is not simply a switch of financing channels. It is an overall shift in the capital perception of Chinese technology assets.
In the past, the overseas market provided a premium. Today, the A-share market provides strategic status, long-term funds, national narratives, and industrial support. The former is a liquidity premium, and the latter is an identity premium. It took Chinese technology companies ten years to realize that the latter is more valuable.
The deeper industrial essence is that there is a natural genetic gap between AI companies and the previous generation of Internet companies.
The core assets of Internet companies are users and traffic, which is a business model that can be understood in any capital market. Facebook can explain it clearly, and so can Tencent. The valuation logic is globally applicable. AI companies are different. Their core assets are computing power, data, and model weights, each of which is deeply bound to national boundaries. Such companies are naturally more suitable to be priced in the domestic capital market.
Zhipu and MiniMax are not the first to realize this. They are the first to take action.
Moreover, their business structures are completely different.
70% of MiniMax's revenue comes from overseas. Its main products are Hailuo AI and Talkie for C-end users, with 236 million users spread across more than 200 countries. It is a Chinese AI company that most resembles a global company.
In contrast, localized deployment has long been Zhipu's primary source of revenue, still contributing more than 70% of its revenue in 2025. Its main customers are in highly regulated industries such as finance, government affairs, and energy, almost all of which are in the domestic market. It is a Chinese AI company that most resembles a national team company.
One is the most global, and the other is the most local.
According to the old financing logic, MiniMax should have focused on the US or Hong Kong stock markets, while Zhipu should have queued up for the A-share market from the beginning.
But by May 2026, these two arch-rivals with completely different genes met on the same track at the crossroads of capital.
Behind them, the return of Yuezhianmian, Lingyiwanwu, Jieyuexingchen, and Baichuan Intelligence to the A-share market is just a certainty that will happen according to the timeline.
An even deeper signal comes from the capital market itself.
The Hang Seng AH Premium Index has dropped from a long-term high of over 130 points to 118.52. Some leading hard-tech H-shares have even seen a rare price inversion between A and H shares. Lanqi Technology, Contemporary Amperex Technology Co., Limited, and GigaDevice Semiconductor are all on this list.
Behind the cold data lies a key signal. After two years of competition, global capital is willing for the first time to give a higher valuation and pay an excess premium for top domestic hard-tech assets in the Hong Kong Stock Exchange market outside the Chinese A-share market.
The two capital markets are no longer shadows or discounted versions of each other. They each have their own responsibilities and bear different strategic foundations. This is the underlying bonus that allows Chinese large model companies to benefit from both sides.
A new capital land bridge has been forcibly built by this generation.
[Beyond the Page] Words:
Ten years ago, the most outstanding Chinese Internet elites were eager to squeeze into New York and Nasdaq.
Today, China's most hardcore AI giants are queuing up to return to the A-share market in an orderly manner.
Nasdaq once defined the golden decade of the previous generation of Chinese Internet companies with US dollars and traffic.
And the A-share market is redefining the endgame of the next generation of Chinese AI large models with national will and industrial infrastructure.
What's really worth watching about MiniMax and Zhipu is not their stock prices.
What's truly thrilling is that at this critical juncture of the great change, they have both made the only necessary choice: choose the A-share market.
This article is from the WeChat official account "Beyond the Page", author: Huahua. Republished by 36Kr with permission.