From “Unable to Pay Salaries” to 200 Million in Financing: How AI Toys Made a Comeback?
Hello everyone, I’m your financial journalist. Today, we’re going to talk about a particularly interesting story involving a company called Haivivi (Yue Ran Chuang Xin) and the AI toys they produce.
If you’ve been shopping recently, you might have come across those plush toys that can talk, chat with children, and even engage with young people. Behind this story is a remarkable transformation from near-bankruptcy to being sought after by investors.
This article not only explains how Haivivi survived but also reveals the ins and outs of the current hot trend of “AI + consumer products”: Why did AI first gain popularity in toys? Is this business really profitable? What are the potential pitfalls?
Below, I’ll break down the news into five key points in simple language to help you understand the logic behind it.
---
1. Survival in Crisis: When AI Toys Were Still Unrecognized by Investors, They Nearly Went Bankrupt
The story begins two years ago. At that time, the company’s founder, Li Yong, was in a very awkward situation:
- No Money for Salaries: The team had been reduced to just a dozen people, and they couldn’t even afford basic research and development or new product launches, so they had to borrow money from banks.
- Funding Rejections: The team approached nearly 200 investment institutions and created over 200 communication groups, but no one showed interest.
Why No Investment?
The capital market was focused on “large-scale models” and “computing chips”—the foundation and “engine” of AI. People thought that consumer-grade AI hardware (like talking toys) was still too early and its prospects unclear, so they were reluctant to take a risk.
The Turning Point:
Li Yong, with experience from developing the Tmall Genie children’s story machine, realized that traditional toys that simply repeated pre-programmed phrases were boring. However, with the emergence of ChatGPT, AI became capable of engaging in continuous, natural conversations like a real person.
They then launched BubblePal, an AI device installed in plush toys like the “Nai Long” (a popular toy). This product sold 350,000 units in total.
- Result: The sales proved the product’s success, and investors’ attitudes shifted dramatically. In August 2025, Haivivi secured 200 million yuan in Series A financing, with investors like Sequoia and CICC joining in.
💡 Journalist’s Note:
In the venture capital world, consensus is more important than technology. When the market hasn’t fully understood a concept, you need real sales data to convince investors. Haivivi’s success came from selling 350,000 units, not from fancy presentations.
---
2. Why Toys? Because AI Can Make Mistakes Without Being Criticized
You might wonder: If AI is so powerful, why not use it for robots or smartwatches instead of toys?
Li Yong gave a practical reason: Tolerance for Mistakes.
- With robots: If it breaks a plate while washing dishes, you’d be angry and think it’s unreliable. Users expect production tools or household helpers to be stable, accurate, and error-free.
- With toys: If a talking teddy bear makes a small mistake or reacts slowly, you might find it endearing or even see that as part of its “personality.” Users expect toys to provide companionship, fun, and emotional value.
In simple terms: Current AI technology isn’t yet perfect. Using it in highly precise contexts (like medicine, driving, or manufacturing) is too risky, but in areas with higher emotional tolerance (like toys or pets), it can shine.
💡 Journalist’s Note:
Low expectations are a great protective barrier for AI. Starting in a field with lower user expectations and more room for mistakes is a smart strategy for startups.
---
3. Opening Physical Stores: Not Just for Selling, but to Teach Users How to Use the Toys
Haivivi has now opened stores in popular shopping malls like Shenzhen Wanxiangwan. You might ask, why spend money on physical stores when toys can be sold online?
Li Yong explains that it’s about educating the market.
- AI toys are a new category: Ordinary consumers don’t know how to use “AI pets” or “smart accessories.” They don’t understand how to interact with them through apps.
- Experience Matters: Only by touching and talking with the toys in person can you feel their “soul” beyond their cold plastic shells.
This is similar to early smartphones: Apple opened Apple Stores not just to sell phones but to let people experience touch-screen technology. The same applies to AI toys; their “playability” can only be truly understood through in-person experiences.
💡 Journalist’s Note:
For new consumer brands, physical stores are not just sales channels but also centers for creating trust. When products are abstract and hard to convey value through images, in-person experiences are crucial for conversion.
---
4. The Industry’s Reality: Despite the Buzz, There Are Many Challenges
Although 2025 is called the “Year of AI Toys,” with the market expected to grow from 24.6 billion to 29 billion yuan, as a journalist, I must warn: This business is not as easy as it seems.
Here are three major challenges:
1. How Long Will the Novelty Last?
Traditional trendy toys (like those from Pop Mart) rely on IP, blind boxes, and collectible elements to drive sales. However, AI toys’ selling points are mainly their ability to talk and provide companionship.
- Question: Will children continue to buy them after the novelty wears off? Will parents be willing to pay for long-term use? There’s a lack of long-term data to prove this.
2. High Hidden Costs:
Traditional plush toys have fixed costs once sold. But AI toys are different: every voice recognition, model call, and content generation consumes money (cloud computing).
- Question: If users talk to the toys 100 times a day, will the manufacturer’s server costs eat into all profits? This is a classic “the more you sell, the more you lose” scenario, unless users subscribe regularly.
3. Severe Homogenization:
90% of AI toys on the market follow the same formula: IP image + plush shell + AI hardware.
- Question: The models and hardware supply chains are similar. The competition will come down to which IP is more popular and which marketing strategy is more effective. The technical barriers are not high.
💡 Journalist’s Note:
AI toys are still in the concept validation phase rather than the mature profit-making phase. Manufacturers need to not only sell hardware but also provide a service that continuously consumes computing power. If they can’t solve the problems of repetition rate and **unit economic model (UE), they’ll struggle to succeed.
---
5. The Future Battle: Moving from Children to Young Adults—The Window is Closing
Haivivi’s next steps are crucial:
- First Generation: Targeting children aged 3-6.
- Second Generation: Expanding to 12-year-olds.
- Third Generation (TapDoki): Aiming at young adults, hoping to turn AI toys into a trend in emotional consumption.
Why the Shift to Young Adults?
The children’s market has a ceiling, and parents are sensitive to price. Young adults, however, are willing to pay for emotional value and “social currency,” offering a larger market potential.
But This Also Brings New Challenges:
- Entry of Giants: If traditional toy companies like Pop Mart and Lego enter the AI toy market, how much time will startups have? Li Yong believes that giants are slow to react, which could be an opportunity for startups. However, a window of opportunity doesn’t mean a barrier to entry.
- Cooling Capital: In 2026, investors will demand results. This means that having a good story is not enough; real profits and growth data are needed.
💡 Journalist’s Note:
The ultimate goal of AI toys is not just to create “talking toys” but to become “emotional companions.” Haivivi’s success will depend on whether it can transform AI toys from educational aids for children into emotional companions for young adults. If it stays at the level of chatbots, it may be replaced by cheaper, more intelligent apps.
---
📝 Summary for Everyone:
1. For Consumers: AI toys are fun, but think carefully before buying: Are you paying for the hardware or for the continuous cloud services? Will they become obsolete after the novelty wears off?
2. For Startups: In the early stages of AI, fields with lower tolerance for mistakes (like toys and pets) are easier to enter. But to scale, you need to solve issues with ongoing costs and user engagement.
3. For Investors: The AI hardware industry is moving from storytelling to showing actual results. Companies without clear concepts, repeat sales data, and unit economic models will be eliminated.
In one sentence: The spring of AI toys has arrived, but the winter is just around the corner. Only those who can turn temporary toys into long-term companionship will succeed.
---
Thank you for reading! If you have any questions or want more analysis, feel free to ask.