Hello, I'm your financial analysis assistant. This article about Zhipu AI and Yushu Technology reveals the most brutal and realistic side of the current AI industry: going public is not a "graduation," but rather an accelerator for burning money.
Many people might see figures like "33.5 billion in financing" or "a plummeting stock price" and think it's just about numbers. But as an economist, I want to tell you that this reflects a life-and-death struggle for AI companies, as they transition from telling stories to engaging in direct competition. Let me break down this news into five key aspects in plain language to help you fully understand the underlying logic.
1. Why do companies still borrow money even after making profits? Because "expenditures exceeding income" is the norm
First, we need to dispel a misconception: just because a company goes public or its revenue increases, it doesn't mean it has no financial problems. On the contrary, although Zhipu AI's revenue has soared by nearly four times (from less than 240 million to 954 million), it has also lost a significant amount of money—2.072 billion yuan in half a year.
It's like a restaurant owner whose daily sales have increased several times, but the money spent on developing new dishes, buying more expensive ingredients, and hiring better chefs is six times the revenue.
- Core logic: The development of large AI models is an endless cycle of spending. Zhipu AI invests 70%-80% of its R&D budget in computing power (i.e., buying the most expensive graphics card servers).
- Reality: As long as the gap between R&D spending and actual revenue (the "scissors gap") remains, financing is essential. Going public only makes borrowing money easier and cheaper, not eliminating the need for it. Of the 33.5 billion yuan Zhipu raised, 60% will be used for the next generation of models, indicating it's still in the starting phase of a competitive race and has a long way to go before making a profit.
2. Yushu Technology's rollercoaster: The market expects a future value, but the company faces immediate challenges
Yushu Technology, which makes robots, had an even more dramatic story. Its IPO price-to-earnings ratio was as high as 219 times (the industry average is 38 times), suggesting investors believed it would earn a lot in the future and paid for that potential in advance. However, its stock price dropped by 55% in a month, resulting in a market value loss of 250 billion yuan.
Why such a sharp drop? Because expectations didn't match reality.
- Financial facts: Although Yushu's revenue increased by 48%, its profits decreased by 19%. This is because R&D costs soared, and sales expenses doubled. In short, its expenses grew faster than its revenue.
- Business weaknesses: Fund managers pointed out that Yushu's hardware is strong, but its algorithms are still weak, and it hasn't yet established a profitable software subscription model.
- Money usage: Nearly half of the funds raised (2 billion yuan) went into software development. This means a company that sells hardware focuses on software development rather than selling more robots. If the revenue from hardware isn't enough to cover software costs, the stock price will inevitably fall.
3. Robots are harder to make money from than large models: expensive data and fragmented scenarios
Many think that since large models (like ChatG) are so costly, robots should be similar. This is a misconception. The development and costs of embodied intelligence (robots) are more rigid and complex.
- Data challenges: Large models can be trained with massive amounts of data from the internet, which is relatively cheap. Robots, however, need real-world data from actual use cases.
- High costs: Collecting an hour of useful data costs around 1,000 yuan. Training a useful model might require millions of hours of data, costing hundreds of millions. This doesn't even include the energy and hardware costs.
- Fragmented nature: Each new scenario requires re-collecting data and re-training, a cost that large models don't face.
4. The post-IPO survival battle: Large models compete for speed, robots for survival
Although both types of companies are burning money, the nature of their spending determines their post-IPO fate.
- Large models (Zhipu): They compete for generational leadership. By having more computing power, their API revenue can grow exponentially (as evidenced by Zhipu's 27-fold increase in API revenue).
- Financing as a competitive tool: Going public gives them more leverage to win this race.
- Robots (Yushu): They struggle to establish a profitable business model. Hardware margins are low, and software hasn't yet generated revenue.
- Financing as a consumption battle: Every hour of data collection and each iteration costs money with no immediate return.
- Key difference: Large models go public to move faster; robots go public to survive. If technology doesn't meet market expectations, stock prices will plummet.
5. Regulatory and market changes: No more room for just telling stories to raise money
There are also macro-level changes: regulators and the capital market are becoming more cautious.
- Stricter reviews: IPOs for humanoid robots are being more stringent, requiring evidence of sustainable revenue and declining losses. This means that just having a good presentation and a concept is no longer enough to go public.
- Valuation adjustments: Yushu's stock crash is a market correction for overvalued startups. Previously, companies with an AI/robot focus got high valuations, but now investors look at fundamentals: can the profits cover costs, and how strong are the technical barriers?
- Conclusion: For AI companies, going public means higher leverage and the ability to raise more funds. Zhipu's financing and Yushu's stock drop illustrate one thing: in the second half of the AI era, there are no easy wins. You either have the ability to continuously burn money (like Zhipu) or generate stable revenue (which few companies yet do).
Implications for investors:
If you invest in the AI sector, don't just look at revenue growth; also consider the R&D expense ratio and **cash flow.* If a company's revenue is rising rapidly but losses are increasing, and it relies heavily on financing, its stock price will be volatile. The real winners will be those that can convert spending into stable revenue, and there are still few such companies.
In summary, the AI industry is a competitive and costly one. Investors need to focus on more than just revenue growth; they must also assess a company's ability to manage expenses and generate stable income.