AsianFin -- Chinese AI toy pioneer Haivivi has secured US$28 million in a Series A funding round, positioning itself at the forefront of the emerging AI-driven play industry.
The funding, led by CICC Capital, Sequoia China, Huashan Capital, and Joy Capital, will support Haivivi’s expansion into new product lines, talent acquisition, and market growth, as the startup seeks to redefine the emotional connection between children and AI-powered toys.
At this year’s CES, AI toys captured global attention, and Haivivi—one of the first startups to integrate large language models into plush toys—emerged as a standout player. The company’s first product, BubblePal, has sold over 200,000 units since its June 2024 launch, demonstrating the market’s appetite for emotionally engaging AI experiences.
“Our investors are more than financial backers—they are strategic partners aligned with our vision,” said Haivivi founder Li Yong. “The AI toy industry is just getting started, and we are acutely aware of the challenges ahead.”
Li, a former founding member of Tmall Genie, brings extensive tech experience to the venture. Reflecting on Haivivi’s early days, he emphasized the importance of avoiding head-to-head competition with tech giants. “Whether it’s funding, talent, or distribution channels, there’s simply no way to go directly against the big players,” he said. According to Li, while major tech firms prioritize markets like AI smartphones and glasses, plush AI toys offer a unique opportunity to deliver emotional value—an area largely overlooked by larger competitors.
Haivivi’s early journey was far from easy. A year or two ago, the company struggled to raise funds, forcing Li to borrow money to cover salaries and even consider mortgaging his house. “We were on the brink of liquidation,” he said. “Fortunately, with support from Professor Gao Bingqiang and introductions to key industry partners, we managed to persevere.” Professor Gao, Haivivi’s angel investor, continues to provide financial backing and strategic guidance, helping the startup maintain its lead in product innovation and IP partnerships.
In June 2024, Haivivi launched its first AI toy, BubblePal, featuring role-playing, multilingual conversations, AI-powered collaborative storytelling, and long-term memory. In the first month, more than 10,000 units sold; total sales now exceed 200,000 units. The success has validated investor confidence in the AI toy market.
Haivivi has focused on intellectual property (IP) as a competitive edge. The company has secured licenses for popular IPs such as Ultraman, Peppa Pig, and Milk Dragon, while also co-creating original IPs with artists. “IP is the soul of a toy,” Li said. “We combine top-tier licensing partnerships with artist collaborations to create and operate engaging new IPs.”
Building on this strategy, Haivivi is set to release its second-generation AI toy, CocoMate, through Leap Innovation. The CocoMate series features a detachable AI module embedded within plush toys, offering more flexible and human-like emotional interaction. With capabilities like multiple wake words, themed card games, and chatroom modes, the initial launch will include a limited-edition Ultraman version and Haivivi’s proprietary “Bubble” gift box, going on sale August 27.
Unlike products that overload features, Li emphasizes simplification. “Integration with smart home systems is trivial, but we deliberately avoid it. We want the toy to be a friend, not an assistant. Too many functions dilute emotional value.”
Industry analysts note the AI toy market is entering a phase of rapid expansion, driven by AI evolution and rising demand for emotionally engaging products. Hardware costs for AI toys are relatively low, making replication by manufacturers such as those in Huaqiangbei feasible. However, Li is unfazed by low-price competitors. “Manufacturing is the easy part. Selling the product successfully is far more challenging,” he said.
Haivivi positions itself as a market leader with strong IP and user engagement as its competitive moat. “In the short term, our IP is our barrier; long term, it may be the memory and loyalty of our users. There’s still a lot of exploratory work ahead,” Li added.
