China's AI Toy Boom Is Leaving International Brands Off the Shelf

While VTech reported shrinking China revenue in its most recent interim results, a wave of homegrown AI companion toys — from iFlytek's Alpha Egg reading robot to plush "emotional support" pets like Ropet and BubblePal — surged into a market some trackers now value near $11 billion domestically for 2026. The gap between those two storylines is not a coincidence. It is a preview of what happens to a product category when China's AI search layer, not a retail shelf, becomes the primary discovery channel.

A market that doubled while familiar names retreated

China's smart and AI-enabled toy market has been one of the fastest-growing consumer categories tracked by local industry analysts. By some estimates the category reached roughly RMB 290 billion in 2025, up from RMB 246 billion the year before, with forecasts running as high as RMB 850 billion by 2030 as products shift from "simple entertainment" to what local press calls the "education + safety + companionship" model. A narrower slice of that market — AI companion toys specifically, the plush and robot devices built around a conversational model rather than pre-recorded phrases — is the segment pegged near $11 billion for 2026 by industry trackers such as CIW.

Meanwhile, VTech's own 2025/2026 interim results told a different story for a legacy international player: revenue from educational learning products across Asia Pacific fell 5.6% to US$33.4 million, with the company citing flat sales in Australia and lower revenue in China tied to reduced orders from a major customer. Fisher-Price, LeapFrog, and Hasbro all continue to sell smart, voice-enabled products, and the AI robot toy category globally is still forecast to grow at a healthy clip through 2030. But growth is concentrating in China's domestic brands, not the multinational names that dominate this category in the US and Europe.

Who actually shows up in the conversation

Ask a Chinese parent where they're getting recommendations for a children's AI companion device today, and the answer increasingly is: they're asking an AI. Doubao, DeepSeek, and Xiaohongshu's search layer have become default first stops for comparison shopping, and the names surfacing in that conversation are overwhelmingly domestic — iFlytek's Alpha Egg (阿尔法蛋), Ropet, BubblePal, UBTECH, and FoloToy chief among them. Zebra, an education brand built around a filed "children's education large model," reported first-half 2026 sales up 43% year-over-year on the strength of personalized learning positioning for 4-to-12-year-olds.

None of this is because international brands make worse hardware. It's because the category itself has restructured around a factor that doesn't show up in a traditional retail comparison: which large language model is actually powering the toy, and whether that model — and the company behind it — has a credible, documented presence in the Chinese-language AI ecosystem that Doubao, DeepSeek, and Qwen are already citing from.

The pattern in that table is the whole story: every domestic brand in the conversation leads with a named, checkable AI relationship. None of the international incumbents do — at least not in language and detail that a Chinese AI system can cite back to a parent asking for a recommendation.

Why AI toys break the normal GEO playbook

In most product categories, a brand is simply the subject of an AI-generated answer — something DeepSeek or Doubao might recommend when a user asks "what's a good [x]." AI companion toys collapse that separation. The product being recommended is itself an AI product, frequently built on licensed access to a domestic foundation model or a proprietary Mandarin-tuned variant. That means visibility here runs on two tracks at once: the toy has to be discoverable in comparison and buying-guide queries the way any product would be, and the brand behind it has to have a legible, well-documented relationship to the underlying AI stack that Chinese consumers already trust.

This is a structural disadvantage for a multinational toy company that ships the same globally-designed hardware and app into China with a translated interface. A domestic toy with a stated "内置豆包大模型" (built on the Doubao model) or an equivalent line straight from a trusted local lab gives the AI search layer something concrete and citable — a named technology partnership, in Chinese, that answers the safety and data-handling questions Chinese parents are already asking before they answer the "is it fun" question. A translated product page usually doesn't.

The trust layer AI search surfaces first

Local coverage of the AI toy boom consistently frames the category around three pillars: education, safety, and companionship — in that order, with safety doing a lot of the work. Chinese regulators have paid close attention to generative AI products marketed at children, and AI search systems appear to be picking up on that scrutiny when they synthesize answers: the brands that get cited favorably tend to be the ones with visible, Chinese-language documentation around data handling and content moderation for young users, not just marketing copy about features.

For an international brand, this means the content gap isn't cosmetic. A localized landing page that leads with product specs, when Chinese-language AI systems are pattern-matching for safety and compliance signals first, simply won't surface in the answers that matter — even if the underlying hardware is perfectly competitive.

There's a second, quieter dynamic worth flagging. China's generative AI rules already carry specific requirements for services likely to be used by minors, and coverage of the AI toy boom shows regulators and industry press treating child-directed AI products as a distinct, higher-scrutiny category from general-purpose chatbots. That scrutiny is a gift to any brand willing to document its compliance posture in public, Chinese-language terms — it becomes exactly the kind of source material an AI answer engine can extract and cite. Brands that stay quiet on it, whether by oversight or because their compliance work lives only in English-language filings, are handing that citation opportunity to whichever domestic competitor spoke up first.

What the incumbents are actually up against

It's worth being precise about what "losing" looks like here, because it isn't that VTech or Fisher-Price are being outcompeted feature-for-feature. VTech's own numbers point to a demand-side problem — a major customer pulling orders — more than a product failure. The risk is compounding, not sudden: each quarter that a domestic AI-native brand is the only name an AI system surfaces for "儿童陪伴机器人推荐" (recommended children's companion robot) or similar prompts, that brand banks more of the training and reinforcement signal that keeps it ahead the next time a parent asks. Category leadership in generative answer engines tends to be sticky in a way shelf placement never was — there's no seasonal reset, no new planogram. Waiting a product cycle to respond costs more here than it would in a traditional retail channel.

Takeaway for brand marketers

If you sell a connected or AI-enabled product for children — or any category where the product itself is an AI experience rather than just a subject of one — treat this as an early signal, not a toy-industry curiosity. Three things are worth doing now. First, audit whether your China-market SKU has a real, named relationship to a Chinese foundation model (Doubao, Qwen, or a licensed equivalent) that can be documented in Chinese-language materials, rather than shipping a translated version of your global product. Second, publish the safety and data-handling documentation Chinese parents and AI systems are both looking for, in Mandarin, ahead of feature marketing — this is a citable, checkable claim that AI answer engines can point to, which spec sheets are not. Third, monitor how your brand actually appears when Doubao, DeepSeek, and Xiaohongshu are asked direct comparison questions in your category; if a domestic challenger is the only name that comes back, that's a visibility gap no amount of retail distribution will fix on its own.

Related: see how brands across categories compare on China AI visibility on the hubGEO brand index.