Summary of Key Points
As the “first stock in the global large-model market,” Zhipu’s stock price has recently experienced a dramatic V-shaped reversal (a sharp drop followed by a violent rebound). This is not due to a sudden change in its fundamental fundamentals, but rather emotional fluctuations triggered by share releases and massive placements. The company, which was once valued at over HK$1.3 trillion based on its AI capabilities, is now facing three major challenges:
1. Severe cash shortage: It is burning through funds at a rate of over HK$20 million per day, and the capital raised from its IPO is quickly depleting.
2. Losing its technological lead: Its competitive edge has been surpassed by rivals, such as Moon’s Dark Side’s Kimi K3, which outperforms Zhipu’s GLM-5.2 in terms of programming capabilities and pricing.
3. Annual stable revenue (ARR) of $1 billion: While this figure seems impressive, its gross margin is declining, and it lacks C-end users, making profitability a distant goal.
4. Strategic shift to AGI (Artificial General Intelligence): Zhipu’s attempt to move towards AGI has raised concerns from the market about its ability to keep up due to a lack of a solid business foundation.
1. Stock Price Volatility: Not a Sign of Poor Performance, but a Result of Financial Issues
The sharp fluctuations in Zhipu’s stock price are not caused by major problems with its business operations. Instead, they are due to the following factors:
- Limited tradable shares: Six months after its listing, only 2.67% of the total shares are available for free trading, which means even small purchases or sales can significantly impact the stock price.
- Share releases: On July 8th, shares held by 11 early investors (cornerstone investors) were released, amounting to more than twice the number of tradable shares. Despite their claims of long-term commitment, this caused panic in the market.
- Desperate need for funds: The day after the share releases, Zhipu announced a placement of HK$314 million, citing the high cost of AI computing power. In fact, 93% of the IPO funds had already been spent, and without additional funding, the company would have run out of cash—this is the core reason for the stock price drop.
2. Losing Its Technological Lead
Zhipu once led the industry with its GLM series of models, but now it has been overtaken by competitors:
- Rival breakthroughs: Zhipu released GLM-5.2 in June, claiming to be the strongest code model in the world. However, a month later, Moon’s Dark Side’s Kimi K3 emerged with 3 trillion parameters (the largest open-source model to date), outperforming GLM-5.2 in programming capabilities and winning first place in user-voted “large-model competitions.”
- Declining technical recognition: The price difference between the two models reflects a shift in market perception: Kimi K3 is considered more valuable, indicating that Zhipu has lost its technological edge.
- Loss of uniqueness: Zhipu’s status as the “first large-model stock” no longer holds significant value, as both Moon’s Dark Side and DeepSeek are preparing to go public, eroding its competitive advantage.
3. $1 Billion in Revenue: An Illusory Bright Spot
Zhipu claims to have annual stable revenue of $1 billion, a 15-fold increase, but upon closer inspection, there are many issues:
- Declining margins: In 2025, its revenue was HK$724 million, with a net loss of HK$4.7 billion, and its gross margin dropped from 56% to 41%. This is because it relies heavily on B-end enterprise customers (73.7% of its revenue), and the more these customers use its API, the higher its computing costs—meaning the more it earns, the more it spends on computing power, resulting in larger losses.
- Lack of C-end users: AI-native apps have nearly 500 million monthly active users, with companies like DouBao, QianWen, and DeepSeek ranking among the top three. Zhipu is not even in the top ten, and a lack of C-end users means it misses out on valuable user data for model training, which will slow down its progress.
- Vulnerable revenue structure: B-end customers can switch easily if competitors offer better technology or lower prices, leading to significant losses.
4. The “Reach for the Sky” Strategy: Vision or a delaying tactic?
To overcome these challenges, Zhipu has announced a strategy focused on AGI, aiming to avoid short-term profitability. Is this approach credible?
- The challenge of a research-oriented background: Zhipu, with its roots in Tsinghua University’s laboratory, excels in fundamental research but struggles with productization. The industry is moving towards commercialization, and Zhipu’s focus on AGI may be counterproductive.
- Market skepticism: With its current financial difficulties, it’s uncertain whether it can afford the additional costs associated with AGI development. Acquiring computing power companies and building data centers may help, but whether this will be enough to catch up with competitors is uncertain.
- Maintaining valuation: The AGI narrative may help maintain investors’ interest, but without stable revenue, how long can this story hold?
Conclusion
Zhipu’s challenges reflect the transition of the AI industry from a phase focused on hype and imagination to one that demands tangible financial performance. Companies that once valued based on scarcity and potential are now forced to demonstrate real profitability. For Zhipu, it must either quickly expand its C-end user base and reduce costs or make breakthroughs in AGI. Otherwise, its status as the “first large-model stock” could soon fade.
(The entire analysis is presented in plain language, avoiding technical jargon, and each point is illustrated with concrete examples to make it understandable to non-financial readers.)