NextFin News - MiniMax Group has quietly become the single most-traded stock among mainland China investors flowing into Hong Kong, edging past Alibaba and Tencent for turnover in the southbound Stock Connect channel. On Sept. 1, 2026, the Shanghai-based artificial-intelligence startup changed hands in HK$6.97 billion of mainland buying and selling — the largest southbound turnover of any Hong Kong-listed name that day, ahead of Zhipu AI at HK$6.25 billion and Alibaba-W at HK$5.02 billion. The shift marks a turning point in how China's household savings are being deployed: no longer into property or bank deposits, but into a loss-making, eight-month-old public AI company that mainland investors are treating as the purest listed bet on the country's artificial-intelligence ambitions.
The Numbers Behind The Shift
The flow data tells a story of rapid migration. MiniMax-W (0100.HK) drew HK$762.6 million of net southbound buying on Sept. 1, even as the Hang Seng Index fell 0.93 percent. Total southbound turnover that day reached HK$92.02 billion, with a net inflow of HK$7.757 billion split between the Shanghai and Shenzhen channels — HK$6.219 billion through Shanghai Connect and HK$1.537 billion through Shenzhen Connect. MiniMax alone accounted for roughly 7.6 percent of the entire day's southbound turnover, a share that would place it among the most-concentrated single-stock flows in the channel's history for a newly listed issuer.
This was not an isolated spike. Two weeks earlier, on Aug. 14, MiniMax already topped the southbound active list with a net buy of HK$1.364 billion, just ahead of Zhipu's HK$1.322 billion. Cumulative southbound net inflows into MiniMax have reached roughly HK$43.4 billion — an amount equal to about 30 days of the stock's average turnover, a concentration that few Hong Kong large-caps sustain outside of index-rebalancing weeks. By comparison, Zhipu AI has drawn cumulative southbound net inflows of roughly HK$43.9 billion, equal to about 23.5 days of its average turnover. The two AI names together now absorb more mainland capital than most of the legacy technology blue chips that have anchored the channel since its 2014 launch.
The mechanics behind the surge are straightforward. MiniMax, which listed on the Hong Kong Stock Exchange on Jan. 9, 2026 at an offer price of HK$165, only recently became eligible for Stock Connect trading. Its inclusion in the Hang Seng Tech Index opened the gateway through which mainland investors — barred from buying most Hong Kong shares directly — can now purchase the company through their Shanghai and Shenzhen brokerage accounts. Zhipu AI, the rival large-language-model developer that listed a day earlier, gained access through the same channel.
The timing matters. MiniMax's debut was one of the largest Hong Kong AI IPOs of the year. The company raised HK$4.8 billion gross, and its shares closed the first day at HK$345, a 109 percent jump from the offer price. That outpaced Zhipu's modest 13 percent first-day gain. Since then the stock has been volatile — touching a high near HK$1,238 before retreating to roughly HK$343 in early September — but the pullback has not deterred mainland buyers. If anything, it has invited them in.
Why Mainland Money, And Why MiniMax
The first question is why mainland investors are choosing MiniMax over the established technology giants that have dominated southbound flows for a decade. The answer lies in scarcity. For mainland retail and private-banking money, the universe of listed pure-play AI companies is tiny. Tencent and Alibaba are diversified conglomerates; their AI businesses are real but buried inside advertising, gaming, payments, and cloud operations. MiniMax is different: it is a single-product, single-theme vehicle. When a mainland investor buys MiniMax, they are making an undiluted bet on Chinese large-language models.
That purity comes with a price. The company's own prospectus shows revenue of US$53.4 million in the nine months ended Sept. 30, 2025, up about 174 percent from a year earlier, alongside a net loss of US$512 million over the same period. It served more than 200 million cumulative users across over 200 countries and regions as of September 2025. In other words, MiniMax is being valued on growth and strategic positioning, not earnings — a framing that mainland investors, long accustomed to pricing stocks on narrative and policy tailwinds, are comfortable with.
The second driver is familiarity. MiniMax's consumer-facing products — chatbots, image generation, and video synthesis — are widely used inside China. Mainland investors interact with the technology directly, which lowers the perceived risk compared with buying a company whose products they never touch. That creates a feedback loop: heavy retail usage builds conviction, conviction drives southbound buying, and rising turnover attracts momentum traders who care less about fundamentals than about where the next wave of money is flowing.
A third factor is the absence of alternatives at home. China's A-share market offers limited exposure to frontier AI models; the country's most prominent model developers have historically been private, backed by venture capital that mainland households cannot easily access. Stock Connect closes that gap. It transforms a Hong Kong IPO into something functionally equivalent to an A-share listing for onshore money, but without the regulatory friction of an onshore flotation. The result is a new kind of cross-border arbitrage: mainland investors paying Hong Kong prices for assets they view as domestic champions.
Cyclical Wave Or Structural Shift
This is where the distinction between cyclical and structural forces becomes critical, and getting it wrong flips the conclusion. The cyclical leg is easy to identify: MiniMax's inclusion in Stock Connect and the Hang Seng Tech Index created a one-time eligibility event, and the initial rush of buying is partly a positioning adjustment as mainland funds that track or benchmark against these indices add the stock. Momentum traders piling in after a 109 percent first-day pop are also cyclical — they will leave as quickly as they arrived if the price stalls.
The structural leg runs deeper. China's household sector is sitting on record deposits after years of weak property returns and falling yields on wealth-management products. That savings pool needs somewhere to go, and Beijing has been steering it toward "hard technology" — semiconductors, artificial intelligence, advanced manufacturing — as part of a broader push for technological self-reliance. The U.S. export curbs on advanced AI chips have only sharpened that political priority, turning domestic AI champions into policy-favored assets. Stock Connect is the pipe that lets this savings pool reach Hong Kong-listed AI names without leaving the onshore regulatory perimeter.
The two forces are separable but mutually reinforcing. The eligibility event is a one-off; the savings reallocation is a multi-year trend. The evidence leans structural, with a cyclical overlay. The HK$43.4 billion cumulative inflow — roughly 30 days of average turnover — is too large and too sustained to be explained by index-tracking alone. It reflects a genuine portfolio shift by mainland money into a new asset class it previously could not access.
The listing-venue implications extend beyond MiniMax. Hong Kong has spent years trying to convince high-growth Chinese technology companies that a local listing is worth the valuation discount to U.S. peers. The exchange's July 2026 listing-rule reforms lowered the financial thresholds and relaxed voting-rights requirements to attract exactly this kind of issuer.
「此次改革是提升香港上市機制靈活性和多樣性的重要一步，確保我們的上市機制與時俱進，能夠應對日益激烈的國際競爭。」香港交易所上市主管伍潔鏇在諮詢總結中表示。她補充稱，希望通過拓寬公司上市渠道，吸引更多類型的優質公司來香港上市，為投資者創造更多機遇。
Translated, the exchange's listing chief said the reform was an important step in improving the flexibility and diversity of Hong Kong's listing regime to meet increasingly fierce international competition, and that widening the channels for companies to list would attract more quality issuers and create more opportunities for investors. MiniMax is living proof that the strategy can work: a Hong Kong listing now offers something a New York listing cannot in the current geopolitical climate — direct access to China's deepest pool of savings.
The AI Tigers: A Theme Bigger Than One Stock
MiniMax does not stand alone. It is one of China's so-called "AI tigers" — a cohort of well-funded model developers that includes Zhipu AI, Baichuan, and Moonshot AI — racing to close the gap with U.S. leaders such as OpenAI. The emergence of two AI-tiger listings within 24 hours of each other in January 2026, followed by their joint entry into Stock Connect, signals that the theme is now investable at scale rather than as a single-name story.
The pairing creates a new dynamic for mainland portfolios. Investors who want AI exposure but distrust single-company risk can split capital between MiniMax and Zhipu, effectively building a two-stock index of China's model layer. That is precisely what the flow data suggests is happening: on Aug. 14, the two names took the top two spots on the southbound net-buy list, and on Sept. 1 they occupied the top two spots by turnover. The correlation between their flows is the fingerprint of thematic, rather than idiosyncratic, demand.
But the cohort is also a warning. None of the AI tigers has yet established a durable moat. Model quality is improving so rapidly that today's leader can be tomorrow's also-ran, and enterprise contracts in China's AI market remain small and experimental. The risk is not that the theme fails — Beijing's support for domestic AI is a structural given — but that the winners within the theme change faster than public-market investors expect. A stock priced for dominance that ends up merely surviving will be punished severely, regardless of how much policy support the sector receives.
The Counter-Thesis: Momentum Bubble, Not Regime Shift
The strongest argument against the structural read is simple: MiniMax is a loss-making company trading at a valuation that assumes a dominant market position it has not yet earned. Its US$53.4 million in revenue in the nine months ended September 2025 against a US$512 million loss is not the profile of a business that can absorb HK$43 billion of investor capital without a reckoning. The counter-thesis holds that southbound investors are chasing a narrative, not underwriting cash flows, and that when the narrative fades — or when U.S. chip restrictions bite harder than expected — the money will leave as fast as it came.
There is merit to this view. Mainland retail flows have a well-documented history of inflating and then abandoning thematic trades, from new-energy vehicles to metaverse concepts. And MiniMax's competitive position is far from secure: it is one of China's "AI tigers" alongside Zhipu, Baichuan, and Moonshot AI, none of which has yet established a durable moat. If a rival ships a materially better model or wins a flagship enterprise contract, MiniMax's premium could compress quickly. The valuation math underscores the risk: at roughly HK$343 a share, MiniMax's market capitalization stands near HK$120 billion against trailing twelve-month revenue of about HK$165 million — a price-to-sales multiple above 700 times that leaves almost no room for execution errors. For context, that multiple is an order of magnitude above the mature cloud-software names that mainland funds also hold through the channel.
The counter-thesis is strongest on the valuation question, but it underestimates the channel effect. Even if MiniMax the company disappoints, the mechanism — mainland savings seeking listed AI exposure through Stock Connect — does not disappear. The money may rotate to Zhipu or the next AI IPO, but it stays within the theme. That is why the structural call survives even a bearish view of MiniMax's own shares. The theme is the trade; the individual stock is merely its current vessel.
The falsifying signal is specific and observable. If MiniMax's average daily southbound turnover falls below 10 percent of its total turnover for a sustained month while the Hang Seng Tech Index is flat or rising, the "structural reallocation" thesis is wrong and the move was purely eligibility-driven momentum. A second signal would be a sustained net southbound outflow from MiniMax exceeding HK$1 billion over five consecutive trading days without a company-specific negative catalyst.
What Comes Next
The mechanism is clear: Stock Connect eligibility converted MiniMax from an inaccessible Hong Kong IPO into the default listed vehicle for mainland China's AI bet, and HK$43.4 billion of cumulative inflows confirm that the money is real, not speculative noise.
Who benefits and who is exposed is asymmetric. MiniMax itself gains a permanent, deep-pocketed shareholder base that can fund its losses while it scales — a lifeline for a company burning more than US$500 million a year. Zhipu AI, sharing the same channel, benefits from the same theme even when MiniMax-specific news is quiet. Hong Kong's exchange benefits most of all: it has discovered a unique selling point that no other financial center can replicate, and that could reshape where China's next generation of technology companies chooses to list.
The exposed parties are the late-arriving momentum traders. A stock that has already moved from HK$165 to HK$1,238 and back to HK$343 in eight months is not a stable store of value; it is a sentiment gauge. Investors who bought the peak on the assumption that southbound flows only go up are the ones carrying the risk if the channel flips.
The forward look splits by time horizon. In the short term — the next one to three months — MiniMax's price will track southbound flow data more closely than fundamentals. Watch the daily Stock Connect turnover rankings: as long as MiniMax remains in the top three, momentum holds. In the medium term — six to twelve months — the question is commercialization. Can MiniMax convert its 200 million users into revenue growth that narrows the US$512 million loss? The next quarterly filing will be the test. In the long term — beyond a year — the structural thesis stands or falls on whether Beijing keeps the Stock Connect gate open for loss-making AI companies and whether U.S. chip restrictions force a technological decoupling that makes domestic champions irreplaceable.
Three scenarios frame the path ahead. The base case is continued elevated but volatile southbound participation, with MiniMax trading as a high-beta proxy for China AI sentiment. The upside case is a successful product or enterprise win that validates the valuation and pulls the stock back toward its highs. The downside case is a regulatory or competitive shock that triggers sustained southbound outflows and forces a repricing toward fundamentals — a level far below current prices.
MiniMax is not just mainland China's new favorite stock; it is the first real test of whether Hong Kong can become the listing home for China's AI ambitions. The money has voted. The company now has to prove it is worth the bet.
Explore more exclusive insights at nextfin.ai.
