AI Investment Trends: From Blindly Chasing High Prices to Rational Screening – What’s Happening in the Primary Market?
Hello everyone, I’m your financial journalist. Recently, there’s been a lot of discussion about AI (Artificial Intelligence), especially the term “bubble,” which has people feeling anxious. On September 11th, at the Bund Conference, several top investors from leading firms such as Zhongke Chuangxing, Fengrui Capital, and BAI Capital came together to talk about their views on AI, how they invest, and their perspectives on the much-hyped “embodied intelligence” (i.e., robots) bubble.
In short, AI is still the hottest sector, but investors are no longer buying into things blindly. They have become more selective, more specific, and more rational. In the past, they would invest in whatever had the latest concept; now, they focus on those that can actually solve problems and survive as models evolve.
Below, I’ll break down this high-level investor discussion into five key points and explain the ins and outs in plain language.
---
1. The Investment Logic Has Changed: No Longer Just About Concepts, but About Whether Intelligence Is Really Improving
In the past, investing in AI might have involved reviewing a PPT or listening to a story before making a decision. Not anymore. Zhao Fuheng, the investment director at Jiuhe Venture Capital, summarized a very crucial criterion: In a specific context, is the improvement in intelligence the main focus right now?
It’s like going to a restaurant for a meal:
- Scenario A: The chef has just learned a new dish, and it tastes a bit better than yesterday. You can still tell the difference, so you’re willing to pay for this improvement, and investors are willing to invest in the restaurant because its core competitiveness is still growing.
- Scenario B: The chef has already reached perfection, and the taste is so good that you can’t tell any difference, or you think it’s good enough. Spending a lot of money on such a small improvement wouldn’t be worthwhile.
Investors now think this way: If users can clearly see that AI has become smarter, they’ll continue to invest; if users no longer notice any improvement, then the investment logic for that project changes, and it might not be worth investing heavily to further enhance its intelligence. Instead, they should consider how to reduce costs and make it more practical.
---
2. The Lifeline for Application Layer Technologies: Don’t Be Just a Backup for Model Providers
This was one of the most pointed discussions. Many AI application companies (e.g., AI coding, AI-generated videos, AI chatbots) are in an awkward position because model providers like OpenAI, Baidu, and Alibaba can also do the same things, and possibly even better.
Investors from Fengrui Capital and BAI Capital pointed out that if an AI application simply uses large models without any unique features, it can easily be overshadowed by the capabilities of these providers.
So, what kind of AI applications are worth investing in?
- For consumer-facing (To C) products: They must offer new ways of interaction beyond just text-based conversations. For example, using voice, gestures, or even brain-computer interfaces to create an experience that the model alone cannot provide.
- For business-facing (To B) products: The key is whether they solve specific problems and whether they can be retrained based on industry data. In other words, can they make the model more tailored to the industry rather than just using generic models?
In one sentence: If your product is just a “skin” for the model, it’s dangerous; if it’s the “soul” or “extremities” of the model, then it’s valuable.
---
3. The Truth About the “Embodied Intelligence” (Robot) Bubble: Is It a Lack of Funds, or Excessive Valuation?
There’s a lot of debate about whether “embodied intelligence” (human-shaped robots, smart hardware, etc.) is a bubble. Some entrepreneurs think the valuations are too high, while others believe it’s not yet at that stage.
Wang Chao from Zhongke Chuangxing even proposed an counterintuitive view: Overall, there’s still not enough funding in the embodied intelligence industry.
Why?
- Technology Is Not Yet Established: The robotics industry is like the early days of the internet—no one knows for sure which path is right. Some companies are focused on making money, some on creating use cases, and some on building computing power. Since the technological paradigm is not yet clear, everyone needs a lot of money to experiment.
- High Barriers: Making robots truly useful requires breakthroughs in both data and models, which requires substantial funding. The current amount of money might only be enough for the leading players to get started, but not for the entire industry.
So, the so-called “bubble” might be more about a mismatch between valuation and the stage of development. Some companies are just starting out, but their valuations have already reached levels typical of a mature industry. Feng Tian from BAI Capital believes that innovation inevitably comes with bubbles because a large amount of capital and talent are necessary for innovation. However, she also warns entrepreneurs: You can ask for a high valuation, but you need to ask yourself if your team and product can survive the cycle. If the valuation is high but the product doesn’t keep up, it could be self-destructive.
---
4. New Channels for Finding Projects: From Networking Events to Social Media and Hackathons
In the past, investors found projects through industry circles, introductions from acquaintances, and attending various conferences. Now, the barriers to entering the AI startup scene have lowered, and young people can also create impressive technology products, so the channels for discovering talent have changed.
- Social Media: Investors from Fengrui Capital follow platforms like REDnote and overseas social media to find early-stage startup ideas.
- Universities and Hackathons: BAI Capital has set up AI labs to find potential candidates by testing models, tracking research papers, and participating in hackathons.
- Technical Expertise Is More Important: Jiang Haonan from IDG Capital pointed out that AI has lowered the entry barriers for the industry. In the past, startups needed to avoid their weakest areas; now, the focus is on having a strong technical expertise. As long as the founding team has outstanding skills in a specific field and can form a company, they have a chance.
This means that the next unicorn could be founded by someone who’s active on GitHub and shares technical insights on REDnote, rather than by a former executive from a large company.
---
5. Advice for Entrepreneurs: Don’t Be Blinded by High Valuations; Fund According to the Product’s Stage
Finally, several investors unanimously advised entrepreneurs to be more rational:
Jiang Haonan from IDG Capital said that valuations in the primary market have risen significantly, with some companies’ first-round valuations being 3 to 5 times what similar projects were valued two or three years ago. This might seem appealing, but it comes with significant risks:
- High Valuations = High Pressure: If you get a high valuation, investors have high expectations for your growth. If the market fluctuates or your product doesn’t meet expectations, this high valuation can become a ceiling, making it difficult to raise more funds or even leading to liquidation if performance doesn’t meet targets.
- Matching Principle: It’s recommended that startups match the scale and valuation of their funding according to the product’s stage. If your product is still in the early validation phase, don’t chase high valuations. It’s better to have a lower valuation and leave room for growth as the economy goes through cycles.
In summary:
AI investment has not slowed down; it has entered a more challenging “deep water” phase. Investors are no longer investing in ideas but in proven improvements in intelligence and unique product value. For entrepreneurs, now is not the time to blindly pursue high valuations but to refine their products and prove their ability to survive the long term. For the general public, understanding these trends can help you see the true face of the AI industry and avoid being misled by superficial “bubbles.”