Zhipu overtakes MiniMax in Hong Kong AI debut aftermath after May market rerating

The Hong Kong AI stock race started with a valuation punch that looked decisive. Beijing-based Zhipu AI closed its first trading day on January 8 with a market capitalisation of HK$57.9 billion (US$7.4 billion). One day later, Shanghai-based MiniMax listed at HK$106.7 billion, almost twice as large. In other words, the market began with the simpler story: MiniMax looked like the bigger bet.

Five months on, that story has flipped. Zhipu, which trades on the exchange as Knowledge Atlas Technology, has surged ahead, outrunning MiniMax despite the initial head start MiniMax enjoyed right out of the gate. The reversal matters because Hong Kong-listed AI developers are not just competing on models. They are competing on investor expectations, the perceived quality and speed of execution, and how quickly public markets reward whatever investors decide is “next” in AI.

To understand why a listing-day valuation gap can unravel so quickly, it helps to remember how public-market pricing works, especially for fast-moving technology sectors. In early trading windows, investors often anchor on size and headline momentum. MiniMax’s January listing at HK$106.7 billion effectively told the market, “this is the larger platform.” Zhipu’s HK$57.9 billion close on January 8 told a different tale, one that initially put it behind in the scoreboard.

Then the scoreboard stopped being about “who is bigger on day one” and started being about “who can sustain the narrative.” Zhipu’s later surge, even though it began at a lower market cap, suggests that subsequent information and market sentiment pulled capital toward Zhipu. The details in the source are framed around timing and relative performance, but the implication for decision-makers is straightforward: early valuation positioning is fragile in categories where expectations evolve weekly, not yearly.

There is also a listing-name wrinkle that matters for how markets track these companies. Zhipu is traded as Knowledge Atlas Technology. That means the market’s day-to-day attention and analyst coverage may not align neatly with the brand investors associate with the underlying AI lab. When Zhipu surged ahead, it was not simply a “better company wins” story. It was also a demonstration that public-market attention can shift even when the stock’s trading identity is slightly removed from the lab name investors first learned.

On the regulatory and listing backdrop, Hong Kong remains a magnet for investors looking for Chinese tech exposure with a familiar market infrastructure. For Chinese AI developers, the decision to list is part economics, part signaling. A public listing introduces new scrutiny and a new rhythm: quarterly reporting, more immediate market interpretation of performance, and heightened sensitivity to sentiment swings. When two AI developers list within days of each other in January and start with very different valuations, the market essentially sets a baseline for how “risk” is priced across the sector.

That baseline is what got challenged. MiniMax started almost twice as large on its listing day. Then, in the months that followed, Zhipu surged ahead. For boards and executives, this is the second-order lesson: capital can re-rate your company quickly if the market decides you are closer to the next inflection point, or if investors believe your trajectory better matches their expectation of AI monetisation and adoption.

The race also highlights a coordination problem for peers. Once an AI stock begins to pull away, competitors feel it in fundraising expectations, hiring dynamics, and the internal pressure to “show progress now.” Even if a rival team is executing aggressively, the stock market is forward-looking and tends to reward the story it thinks it can sustain. Zhipu’s move ahead after five months is a reminder that “launch order” and “initial market cap” do not determine the end of the story.

The strategic stakes are simple: in Hong Kong’s AI listings, the market can change its mind faster than most board decks are built. For executives at other AI developers, the takeaway is not to obsess over who led on day one. It is to build a durable narrative that can withstand re-rating, because the value gap that looked unbridgeable at launch can vanish, and then reverse, as sentiment and expectations evolve.