Zhipu and MiniMax are becoming more and more like Hermès.

Ask AI · How does Wave Theory explain the pricing logic of AI stock scarcity?

These days, the Hong Kong stock big model sector has once again staged a dizzying “roller coaster” performance.

On April 1st, after Zhipu released its 2025 financial report, the stock price surged about 32% in a single day, with a market capitalization briefly surpassing 410 billion HKD, and MiniMax also rose 14% simultaneously. However, on April 2nd, both companies reversed course and retreated, with Zhipu dropping nearly 15%, and MiniMax falling 10%.

▲Zhipu’s stock price trend after listing on January 8, 2026

If you open the financial report, you’ll find it almost “magical realism”: in 2025, Zhipu and MiniMax’s revenues were only 720 million yuan and 544 million yuan respectively, but their net losses after non-recurring gains and losses reached about 3.8 billion yuan and 1.9 billion yuan. How can such “profitable companies that lose faster” oscillate upward amid doubts? Why, even after a correction, their market value still remains in the 300 billion HKD range, far surpassing many traditional “blue-chip” stocks?

The answer is: these AI newcomers have already broken free from the “valuation gravity field” of traditional manufacturing or internet companies. In the high barrier, high investment, winner-takes-all track of large models, the traditional breakeven line is no longer the only coordinate system.

▲Zhipu’s net profit attributable to parent over the past four years

Market pricing of them is less about paying for current profitability and more about betting on the “infrastructure of the AI era.” This is similar to the internet bubble around 2000—back then, Amazon was losing money year after year, but investors were betting not on its current profit statement, but on the future it represented. The difference is, this story might be faster and more intense than the internet boom.

To understand this “magical performance,” PE ratios or fundamental analysis are no longer enough. Investment theorist Wave proposed a widely cited framework—the “dual commodity attribute of stocks”—which can help us clear some of the fog.

In Wave’s view, stocks carry two identities simultaneously:

First: a normal commodity. Stocks represent a company’s productive capacity, and their “use value”—that is, profitability—determines stock prices. This is the traditional value investor’s perspective.

Second: a scarce commodity. Stocks are trading chips, and their “market price” determines exchange value, which in turn “calibrates” their use value. This is the game theory perspective.

When a stock’s scarcity is fully recognized by the market, its pricing logic undergoes a qualitative change: it is no longer determined by “how much it’s worth,” but by “how much someone is willing to pay.”

Hermès’ Birkin bag is a typical example of “price calibration value”: not something you can buy just because you want it; it requires “waiting list”; buyers’ first reaction is often not “how much can it hold,” but its “value retention and status pricing” in the second-hand market.

Currently, the reality of the Hong Kong stock big model sector is—purely representative of China’s frontier level of general artificial intelligence and capable of large-scale commercialization—extremely scarce. Zhipu and MiniMax are almost the only two options available in the market right now.

For long-term funds and hedge funds in the secondary market, they are no longer just financial statements but an irreplaceable “era chip.” When an asset is extremely scarce, the pricing logic becomes: because everyone is willing to pay a high price, it must be worth that much.

This seemingly crazy pricing logic is not just hype about chips. There is actually a positive feedback mechanism behind it.

When a company’s stock price is driven to an extremely high level, this “premium” translates into real resource allocation rights. Take Zhipu as an example: its current market cap of 350 billion HKD is not just a number on the books—it means the company can raise equity at lower costs, attract top global scientists with richer options packages, and more easily persuade governments and large enterprises to purchase its services.

This is what Wave calls the “discovery of exchange value”: capital markets first give a “sky-high” price through speculation, which grants the company a certain market “legitimacy.” The company then uses the resources gained from the premium to refine products and push the model’s intelligence “upper limit,” solidifying its position as an industry leader.

In short: it’s not because it’s powerful that it’s expensive, but because selling high first provides the “ticket” to become stronger.

The market believes that Zhipu and MiniMax may form a clever closed loop: high R&D investment creates technological scarcity, scarcity attracts market premium, and high market value then feeds back into real R&D. Under this logic, Zhipu is no longer just a company selling API services; it’s more like an “intelligent entity” evolving within the capital market. In this battle for pricing power, traditional accounting standards are no longer keeping pace with AI’s speed.

However, Wave also offers a sober footnote: the profit from investing in scarce commodities is ultimately a game played by a very few.

For ordinary investors, profiting from this game detached from fundamentals faces two harsh physical limits.

Limit one: the maximum investment horizon. The time the stock price diverges from its intrinsic value often exceeds what investors can tolerate in holding periods. A company’s bottoming and “bubble” persistence may surpass most people’s patience. Many exit amid doubts, only to fall prey to the wave of dawn’s early light.

Limit two: the minimum reaction time. The jump in price and the return to value of scarce commodities are often “leap-like.” For stocks like Zhipu, which can rise 30% in a day and fall 15% the next, key corrections often happen within minutes. Without extremely fast reactions and steel-like mental resilience, participants are mostly reduced to “buying high” or “selling low” spectators.

But remember, scarcity is the only support of this logic, and it’s also the reason many high-market-cap companies eventually end up with nothing.

Once the competitive landscape shifts—new strong competitors emerge, diluting the “uniqueness,” or massive R&D investments fail to sustain technological leadership—the “leap-like” return to value may be even more rapid and ruthless than the rise itself.

The pricing power granted by the market is never free. It can elevate an era, but it can also wipe it out in an instant.

From Zhipu to MiniMax, this battle for pricing power is far from over. In the relentless arena of money, every valuation point is a bet. Ordinary people, just watch the show!